Live
Connect
WebSocket
/live/sessionsCreate session
POST
/live/sessionsModels
AudioFormat = AudioPCM { rate, type } or AudioPCMU { rate, type } or AudioPCMA { rate, type } Audio encoding and sample rate for audio sent and received over a Live WebSocket connection. WebRTC and SIP negotiate their media format separately.
AudioFormat = AudioPCM { rate, type } or AudioPCMU { rate, type } or AudioPCMA { rate, type }
Audio encoding and sample rate for audio sent and received over a Live WebSocket connection. WebRTC and SIP negotiate their media format separately.
One of the following: AudioPCM { rate, type } Raw, mono 16-bit little-endian PCM audio for a Live WebSocket connection.
rate: 16000 or 24000 Audio sample rate in hertz. Live WebSocket PCM audio supports 16000 or 24000 Hz.
minimum16000 maximum24000 One of the following: 16000 24000 type: "audio/pcm" (default: "audio/pcm") The audio encoding. Always audio/pcm.
AudioPCMU { rate, type } Raw, mono G.711 μ-law audio for a Live WebSocket connection.
rate: integer Audio sample rate in hertz. G.711 audio uses 8000 Hz.
minimum8000 maximum8000 type: "audio/pcmu" (default: "audio/pcmu") The audio encoding. Always audio/pcmu.
AudioPCMA { rate, type } Raw, mono G.711 A-law audio for a Live WebSocket connection.
rate: integer Audio sample rate in hertz. G.711 audio uses 8000 Hz.
minimum8000 maximum8000 type: "audio/pcma" (default: "audio/pcma") The audio encoding. Always audio/pcma.
ClientDelegation object { type, acknowledgement_delay_ms } Delegate tasks to your application. The Live session emits delegation events that your backend handles.
ClientDelegation object { type, acknowledgement_delay_ms }
Delegate tasks to your application. The Live session emits delegation events that your backend handles.
type: "client" (default: "client")
The delegation owner. Always client for tasks handled by your application.
acknowledgement_delay_ms: optional integer
Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations.
minimum0
maximum800
ClientEvent = SessionStartEvent { session, type, event_id } or SessionUpdateEvent { session, type, event_id } or InputAudioAppendEvent { audio, type, event_id } or 12 moreClient events for Live. Initialize a primary WebSocket with session.start and wait for session.started. WebRTC creation already starts the session. Audio append is primary WebSocket-only. See the Live prompting guide before writing frontend instructions and delegation policies.
ClientEvent = SessionStartEvent { session, type, event_id } or SessionUpdateEvent { session, type, event_id } or InputAudioAppendEvent { audio, type, event_id } or 12 more
Client events for Live. Initialize a primary WebSocket with session.start and wait for session.started. WebRTC creation already starts the session. Audio append is primary WebSocket-only. See the Live prompting guide before writing frontend instructions and delegation policies.
One of the following: SessionStartEvent { session, type, event_id } Start a Live session on a primary WebSocket. Send this event before other commands and wait for session.started.
session: SessionConfig { model, audio, client, 6 more } Initial configuration for a primary WebSocket. Send session.start first and wait for session.started before application commands. WebRTC creation already starts the session; do not send this event again on its data channel.
type: "session.start" (default: "session.start") The Live client event type. Always session.start.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 SessionUpdateEvent { session, type, event_id } Update the delegation settings or transcript triggers of an active Live session. The server acknowledges accepted changes with session.updated.
session: SessionUpdateConfig { delegation, triggers } Sparse delegation and trigger updates. Omitted settings retain their values. The delegation type cannot change. Model, frontend instructions, audio, and startup input are immutable.
type: "session.update" (default: "session.update") The Live client event type. Always session.update.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 InputAudioAppendEvent { audio, type, event_id } Send audio to a Live session over its primary WebSocket. WebRTC and SIP sessions send audio over their media transport.
audio: string Base64-encoded raw audio in the startup-selected format, without a WAV or other container header. Primary WebSocket only; media transports use their audio track. Audio appends have no acknowledgment. Reflected sideband server events reuse this event type and audio key, with no timestamps or event_id; their audio is always mono PCM16LE at 24 kHz.
minLength1 type: "session.input_audio.append" (default: "session.input_audio.append") The Live client event type. Always session.input_audio.append.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 InputAudioMuteEvent { type, event_id } Mute audio input to the Live model without closing the session. The server acknowledges with session.input_audio.muted.
type: "session.input_audio.mute" (default: "session.input_audio.mute") The Live client event type. Always session.input_audio.mute.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 InputAudioUnmuteEvent { type, event_id } Resume audio input to a Live model after muting it. The server acknowledges with session.input_audio.unmuted. An active audio fixture still pauses inference until playback finishes.
type: "session.input_audio.unmute" (default: "session.input_audio.unmute") The Live client event type. Always session.input_audio.unmute.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 OutputAudioFixturePlayEvent { audio, type, event_id } Play one complete audio fixture through the session output. Send this event over the primary WebSocket or a server sideband. Audio must be raw mono PCM16 little-endian at 24 kHz, regardless of the session audio format, and at most 30 seconds. The entire payload is validated before model speech is interrupted and input is muted. Only one fixture can be active. fixture.playing acknowledges acceptance and fixture.played reports server mixer drain. These notifications are not exact transport-media boundaries. Input arriving during playback is dropped and not replayed.
audio: string Base64-encoded raw mono PCM16 audio, little-endian, at 24 kHz. Do not include a WAV or other container header.
minLength1 maxLength1920000 type: "session.output_audio.fixture.play" (default: "session.output_audio.fixture.play") event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 InstructionsAppendEvent { content, delegation_id, type, event_id } Append developer instructions to the Live conversation while it is running, optionally associating them with an existing client delegation. Rejected while an audio fixture is active. These instructions remain in conversation history through compaction and are included in future backend delegations. For external audio, describe content already delivered so the assistant does not repeat it.
content: string Instruction text to append, limited to 500 tokens. This is a plain string, not an array of content parts.
delegation_id: string or null Required, nullable. Set null for general session context, or use the ID from session.delegation.created for an existing client delegation. Non-null IDs are not accepted with Responses delegation.
minLength1 type: "session.instructions.append" (default: "session.instructions.append") The Live client event type. Always session.instructions.append.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 ThinkingAppendEvent { content, delegation_id, type, event_id } Provide silent reasoning or progress context to the Live model, optionally for an existing client delegation. Rejected while an audio fixture is active.
content: string Silent reasoning or progress context, limited to 500 tokens. It does not directly request speech, but can influence later speech and is not a secrecy boundary.
delegation_id: string or null Required, nullable. Set null for general session context, or use the ID from session.delegation.created for an existing client delegation. Non-null IDs are not accepted with Responses delegation.
minLength1 type: "session.thinking.append" (default: "session.thinking.append") The Live client event type. Always session.thinking.append.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 CommentaryAppendEvent { content, delegation_id, type, event_id } Provide context the Live model can communicate to the user, optionally for an existing client delegation. Rejected while an audio fixture is active.
content: string Speakable context for the Live model, limited to 500 tokens. Use this for a result the model should communicate; use session.thinking.append for silent context.
delegation_id: string or null Required, nullable. Set null for general session context, or use the ID from session.delegation.created for an existing client delegation. Non-null IDs are not accepted with Responses delegation.
minLength1 type: "session.commentary.append" (default: "session.commentary.append") The Live client event type. Always session.commentary.append.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 DecisionCreateEvent { event_id, type } Evaluate the configured questions using the current conversation and Decision inputs. Reusing an event_id is rejected for the lifetime of the session.
event_id: string minLength1 maxLength256 type: "decision.create" (default: "decision.create") DecisionItemCreateEvent { item, type, event_id } Append an input item for future decisions without starting an evaluation or changing Live voice context.
item: Message or Message or Message or 5 more One of the following: Message { content, role, id, type } content: string or array of InputText or InputImage or InputJSON The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of InputText or InputImage or InputJSON One of the following: InputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") role: "user" (default: "user") The message role. Always user.
id: optional string or null The unique ID of this message item.
type: optional "message" (default: "message") Message { content, role, id, type } content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "developer" (default: "developer") id: optional string or null The unique ID of this message item.
type: optional "message" (default: "message") Message { content, role, id, type } content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "system" (default: "system") id: optional string or null The unique ID of this message item.
type: optional "message" (default: "message") Message { content, role, id, 2 more } content: string or array of OutputText or Refusal The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of OutputText or Refusal One of the following: OutputText { text, type } text: string The text content.
maxLength10485760 type: "output_text" (default: "output_text") Refusal { refusal, type } refusal: string The refusal text.
maxLength10485760 type: "refusal" (default: "refusal") role: "assistant" (default: "assistant") id: optional string or null The unique ID of this message item.
phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer). For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" type: optional "message" (default: "message") FunctionCall { arguments, call_id, name, 4 more } arguments: string The function arguments as a JSON string.
call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 name: string pattern^[a-zA-Z0-9_.-]+$ minLength1 maxLength128 type: "function_call" (default: "function_call") The item type. Always function_call.
id: optional string or null The local Decision item ID.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" FunctionCallOutput { call_id, output, type, 2 more } call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "function_call_output" (default: "function_call_output") The type of the function tool call output. Always function_call_output.
id: optional string or null The local Decision item ID.
status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCall { call_id, input, name, 4 more } call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 input: string The input to the custom tool call.
name: string The name of the custom tool to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 type: "custom_tool_call" (default: "custom_tool_call") The item type. Always custom_tool_call.
id: optional string or null The local Decision item ID.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCallOutput { call_id, output, type, 2 more } call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "custom_tool_call_output" (default: "custom_tool_call_output") id: optional string or null The local Decision item ID.
status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" type: "decision.item.create" (default: "decision.item.create") event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 DecisionItemDeleteEvent { item_id, type, event_id } Delete an input item from future decisions without starting an evaluation or changing Live voice context.
item_id: string minLength1 type: "decision.item.delete" (default: "decision.item.delete") event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 ResponseItemCreateEvent { item, type, event_id } Add an input item to the Live session’s Responses backend. Requires Responses delegation; use response.create to request a response.
item: EasyInputMessage or Message or ResponseOutputMessage or 30 more An input item to append to the Responses backend conversation, such as a user message or a function tool result.
One of the following: EasyInputMessage { content, role, phase, type } A message input to the model with a role indicating instruction following
hierarchy. Instructions given with the developer or system role take
precedence over instructions given with the user role. Messages with the
assistant role are presumed to have been generated by the model in previous
interactions.
content: string or ResponseInputMessageContentList Text, image, or audio input to the model, used to generate a response.
Can also contain previous assistant responses.
One of the following: TextInput = string A text input to the model.
ResponseInputMessageContentList = array of ResponseInputContent A list of one or many input items to the model, containing different content
types.
One of the following: ResponseInputText { text, type, prompt_cache_breakpoint } A text input to the model.
text: string The text input to the model.
type: "input_text" (default: "input_text") The type of the input item. Always input_text.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputImage { detail, type, file_id, 2 more } An image input to the model. Learn about image inputs.
detail: ImageDetail The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
type: "input_image" (default: "input_image") The type of the input item. Always input_image.
file_id: optional string or null The ID of the file to be sent to the model.
image_url: optional string or null The URL of the image to be sent to the model. A fully qualified URL or base64 encoded image in a data URL.
formaturi prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputFile { type, detail, file_data, 4 more } A file input to the model.
type: "input_file" (default: "input_file") The type of the input item. Always input_file.
detail: optional "auto" or "low" or "high" The detail level of the file to be sent to the model. Use auto to let the system select the detail level; for GPT-5.6 and later models, auto uses high-quality rendering, which may increase input token usage. Use low for lower-cost rendering, or high to render the file at higher quality. Defaults to auto.
One of the following: "auto" "low" "high" file_data: optional string The content of the file to be sent to the model.
file_id: optional string or null The ID of the file to be sent to the model.
file_url: optional string The URL of the file to be sent to the model.
formaturi filename: optional string The name of the file to be sent to the model.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" role: "user" or "assistant" or "system" or "developer" The role of the message input. One of user, assistant, system, or
developer.
One of the following: "user" "assistant" "system" "developer" phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer).
For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend
phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" type: optional "message" The type of the message input. Always message.
Message { content, role, status, type } A message input to the model with a role indicating instruction following
hierarchy. Instructions given with the developer or system role take
precedence over instructions given with the user role.
content: ResponseInputMessageContentList { , , } A list of one or many input items to the model, containing different content
types.
role: "user" or "system" or "developer" The role of the message input. One of user, system, or developer.
One of the following: "user" "system" "developer" status: optional "in_progress" or "completed" or "incomplete" The status of item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" type: optional "message" The type of the message input. Always set to message.
ResponseOutputMessage { id, content, role, 3 more } An output message from the model.
id: string The unique ID of the output message.
content: array of ResponseOutputText or ResponseOutputRefusal The content of the output message.
One of the following: ResponseOutputText { annotations, logprobs, text, type } A text output from the model.
annotations: array of FileCitation or URLCitation or ContainerFileCitation or FilePath The annotations of the text output.
One of the following: FileCitation { file_id, filename, index, type } A citation to a file.
file_id: string The ID of the file.
filename: string The filename of the file cited.
index: integer The index in the output text at which to insert the file citation.
type: "file_citation" (default: "file_citation") The type of the file citation. Always file_citation.
URLCitation { end_index, start_index, title, 2 more } A citation for a web resource used to generate a model response.
end_index: integer The index of the last character of the URL citation in the message.
start_index: integer The index of the first character of the URL citation in the message.
title: string The title of the web resource.
type: "url_citation" (default: "url_citation") The type of the URL citation. Always url_citation.
url: string The URL of the web resource.
formaturi ContainerFileCitation { container_id, end_index, file_id, 3 more } A citation for a container file used to generate a model response.
container_id: string The ID of the container file.
end_index: integer The index of the last character of the container file citation in the message.
file_id: string The ID of the file.
filename: string The filename of the container file cited.
start_index: integer The index of the first character of the container file citation in the message.
type: "container_file_citation" (default: "container_file_citation") The type of the container file citation. Always container_file_citation.
FilePath { file_id, index, type } A path to a file.
file_id: string The ID of the file.
index: integer The index of the file in the list of files.
type: "file_path" The type of the file path. Always file_path.
logprobs: array of { token, bytes, logprob, top_logprobs } token: string bytes: array of number logprob: number top_logprobs: array of { token, bytes, logprob } token: string bytes: array of number logprob: number text: string The text output from the model.
type: "output_text" (default: "output_text") The type of the output text. Always output_text.
ResponseOutputRefusal { refusal, type } A refusal from the model.
refusal: string The refusal explanation from the model.
type: "refusal" (default: "refusal") The type of the refusal. Always refusal.
role: "assistant" The role of the output message. Always assistant.
status: "in_progress" or "completed" or "incomplete" The status of the message input. One of in_progress, completed, or
incomplete. Populated when input items are returned via API.
One of the following: "in_progress" "completed" "incomplete" type: "message" The type of the output message. Always message.
phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer).
For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend
phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" FileSearchCall { id, queries, status, 2 more } The results of a file search tool call. See the
file search guide for more information.
id: string The unique ID of the file search tool call.
queries: array of string The queries used to search for files.
status: "in_progress" or "searching" or "completed" or 2 more The status of the file search tool call. One of in_progress,
searching, incomplete or failed,
One of the following: "in_progress" "searching" "completed" "incomplete" "failed" type: "file_search_call" The type of the file search tool call. Always file_search_call.
results: optional array of object or null The results of the file search tool call.
attributes: optional map[string or number or boolean] or null Set of 16 key-value pairs that can be attached to an object. This can be
useful for storing additional information about the object in a structured
format, and querying for objects via API or the dashboard. Keys are strings
with a maximum length of 64 characters. Values are strings with a maximum
length of 512 characters, booleans, or numbers.
One of the following: string maxLength512 number boolean file_id: optional string The unique ID of the file.
filename: optional string The name of the file.
score: optional number The relevance score of the file - a value between 0 and 1.
formatfloat text: optional string The text that was retrieved from the file.
ComputerCall { id, call_id, pending_safety_checks, 4 more } A tool call to a computer use tool. See the
computer use guide for more information.
id: string The unique ID of the computer call.
call_id: string An identifier used when responding to the tool call with output.
pending_safety_checks: array of { id, code, message } The pending safety checks for the computer call.
id: string The ID of the pending safety check.
code: optional string or null The type of the pending safety check.
message: optional string or null Details about the pending safety check.
status: "in_progress" or "completed" or "incomplete" The status of the item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" type: "computer_call" (default: "computer_call") The type of the computer call. Always computer_call.
action: optional ComputerAction A click action.
actions: optional ComputerActionList { Click, DoubleClick, Drag, 6 more } Flattened batched actions for computer_use. Each action includes an
type discriminator and action-specific fields.
ComputerCallOutput { call_id, output, type, 3 more } The output of a computer tool call.
call_id: string The ID of the computer tool call that produced the output.
minLength1 maxLength64 output: ResponseComputerToolCallOutputScreenshot { type, file_id, image_url } A computer screenshot image used with the computer use tool.
type: "computer_call_output" (default: "computer_call_output") The type of the computer tool call output. Always computer_call_output.
id: optional string or null The ID of the computer tool call output.
acknowledged_safety_checks: optional array of object or null The safety checks reported by the API that have been acknowledged by the developer.
id: string The ID of the pending safety check.
code: optional string or null The type of the pending safety check.
message: optional string or null Details about the pending safety check.
status: optional "in_progress" or "completed" or "incomplete" or null The status of the message input. One of in_progress, completed, or incomplete. Populated when input items are returned via API.
One of the following: "in_progress" "completed" "incomplete" WebSearchCall { id, status, type, action } The results of a web search tool call. See the
web search guide for more information.
id: string The unique ID of the web search tool call.
status: "in_progress" or "searching" or "completed" or 2 more The status of the web search tool call.
One of the following: "in_progress" "searching" "completed" "failed" "incomplete" type: "web_search_call" The type of the web search tool call. Always web_search_call.
action: optional Search or OpenPage or FindInPage An object describing the specific action taken in this web search call.
Includes details on how the model used the web (search, open_page, find_in_page).
One of the following: Search { type, queries, query, sources } Action type “search” - Performs a web search query.
type: "search" The action type.
queries: optional array of string The search queries.
query: optional string Deprecated.
The search query.
sources: optional array of { type, url } The sources used in the search.
type: "url" The type of source. Always url.
url: string The URL of the source.
formaturi OpenPage { type, url } Action type “open_page” - Opens a specific URL from search results.
type: "open_page" The action type.
url: optional string or null The URL opened by the model.
formaturi FindInPage { pattern, type, url } Action type “find_in_page”: Searches for a pattern within a loaded page.
pattern: string The pattern or text to search for within the page.
type: "find_in_page" The action type.
url: optional string or null The URL of the page searched for the pattern, when available.
formaturi FunctionCall { arguments, call_id, name, 6 more } A tool call to run a function. See the
function calling guide for more information.
arguments: string A JSON string of the arguments to pass to the function.
call_id: string The unique ID of the function tool call generated by the model.
name: string The name of the function to run.
type: "function_call" The type of the function tool call. Always function_call.
id: optional string The unique ID of the function tool call.
async: optional boolean Whether the function tool call runs asynchronously.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
type: "program" (default: "program") namespace: optional string The namespace of the function to run.
status: optional "in_progress" or "completed" or "incomplete" The status of the item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" FunctionCallOutput { output, type, id, 5 more } The output of a function tool call.
output: string or array of ResponseInputTextContent or ResponseInputImageContent or ResponseInputFileContent Text, image, or file output of the function tool call.
One of the following: string A JSON string of the output of the function tool call.
maxLength10485760 array of ResponseInputTextContent or ResponseInputImageContent or ResponseInputFileContent An array of content outputs (text, image, file) for the function tool call.
One of the following: ResponseInputTextContent { text, type, prompt_cache_breakpoint } A text input to the model.
text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") The type of the input item. Always input_text.
prompt_cache_breakpoint: optional object or null Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputImageContent { type, detail, file_id, 2 more } An image input to the model. Learn about image inputs
type: "input_image" (default: "input_image") The type of the input item. Always input_image.
detail: optional ImageDetail or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
file_id: optional string or null The ID of the file to be sent to the model.
image_url: optional string or null The URL of the image to be sent to the model. A fully qualified URL or base64 encoded image in a data URL.
maxLength20971520 formaturi prompt_cache_breakpoint: optional object or null Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputFileContent { type, detail, file_data, 4 more } A file input to the model.
type: "input_file" (default: "input_file") The type of the input item. Always input_file.
detail: optional "auto" or "low" or "high" The detail level of the file to be sent to the model. Use auto to let the system select the detail level; for GPT-5.6 and later models, auto uses high-quality rendering, which may increase input token usage. Use low for lower-cost rendering, or high to render the file at higher quality. Defaults to auto.
One of the following: "auto" "low" "high" file_data: optional string or null The base64-encoded data of the file to be sent to the model.
maxLength73400320 file_id: optional string or null The ID of the file to be sent to the model.
file_url: optional string or null The URL of the file to be sent to the model.
formaturi filename: optional string or null The name of the file to be sent to the model.
prompt_cache_breakpoint: optional object or null Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" type: "function_call_output" (default: "function_call_output") The type of the function tool call output. Always function_call_output.
id: optional string or null The unique ID of the function tool call output. Populated when this item is returned via API.
call_id: optional string or null The unique ID of the function tool call generated by the model.
minLength1 maxLength64 caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
name: optional string or null The name of the tool that produced the output.
minLength1 maxLength128 namespace: optional string or null The namespace of the tool that produced the output.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 status: optional "in_progress" or "completed" or "incomplete" or null The status of the item. One of in_progress, completed, or incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" ToolSearchCall { arguments, type, id, 3 more } arguments: unknown The arguments supplied to the tool search call.
type: "tool_search_call" (default: "tool_search_call") The item type. Always tool_search_call.
id: optional string or null The unique ID of this tool search call.
call_id: optional string or null The unique ID of the tool search call generated by the model.
minLength1 maxLength64 execution: optional "server" or "client" Whether tool search was executed by the server or by the client.
One of the following: "server" "client" status: optional "in_progress" or "completed" or "incomplete" or null The status of the tool search call.
One of the following: "in_progress" "completed" "incomplete" ToolSearchOutput { tools, type, id, 3 more } tools: array of Function or FileSearch or Computer or 13 more The loaded tool definitions returned by the tool search output.
One of the following: Function { name, parameters, strict, 6 more } Defines a function in your own code the model can choose to call. Learn more about function calling.
name: string The name of the function to call.
parameters: map[unknown] or null A JSON schema object describing the parameters of the function.
strict: boolean or null Whether strict parameter validation is enforced for this function tool.
type: "function" (default: "function") The type of the function tool. Always function.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
One of the following: "direct" "programmatic" async: optional boolean defer_loading: optional boolean Whether this function is deferred and loaded via tool search.
description: optional string or null A description of the function. Used by the model to determine whether or not to call the function.
output_schema: optional map[unknown] or null A JSON schema object describing the JSON value encoded in string outputs for this function.
FileSearch { type, vector_store_ids, filters, 2 more } A tool that searches for relevant content from uploaded files. Learn more about the file search tool.
type: "file_search" (default: "file_search") The type of the file search tool. Always file_search.
vector_store_ids: array of string The IDs of the vector stores to search.
filters: optional ComparisonFilter or CompoundFilter or null A filter to apply.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } Combine multiple filters using and or or.
filters: array of ComparisonFilter or CompoundFilter Array of filters to combine. Items can be ComparisonFilter or CompoundFilter.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } This type repeats the enclosing schema. type: "and" or "or" Type of operation: and or or.
One of the following: "and" "or" max_num_results: optional integer The maximum number of results to return. This number should be between 1 and 50 inclusive.
ranking_options: optional { hybrid_search, ranker, score_threshold } Ranking options for search.
hybrid_search: optional { embedding_weight, text_weight } Weights that control how reciprocal rank fusion balances semantic embedding matches versus sparse keyword matches when hybrid search is enabled.
embedding_weight: number The weight of the embedding in the reciprocal ranking fusion.
text_weight: number The weight of the text in the reciprocal ranking fusion.
ranker: optional "auto" or "default-2024-11-15" The ranker to use for the file search.
One of the following: "auto" "default-2024-11-15" score_threshold: optional number The score threshold for the file search, a number between 0 and 1. Numbers closer to 1 will attempt to return only the most relevant results, but may return fewer results.
Computer { type } A tool that controls a virtual computer. Learn more about the computer tool.
type: "computer" (default: "computer") The type of the computer tool. Always computer.
ComputerUsePreview { display_height, display_width, environment, type } A tool that controls a virtual computer. Learn more about the computer tool.
display_height: integer The height of the computer display.
display_width: integer The width of the computer display.
environment: "windows" or "mac" or "linux" or 2 more The type of computer environment to control.
One of the following: "windows" "mac" "linux" "ubuntu" "browser" type: "computer_use_preview" (default: "computer_use_preview") The type of the computer use tool. Always computer_use_preview.
WebSearch { type, external_web_access, filters, 2 more } Search the Internet for sources related to the prompt. Learn more about the
web search tool.
type: "web_search" or "web_search_2025_08_26" (default: "web_search") The type of the web search tool. One of web_search or web_search_2025_08_26.
One of the following: "web_search" "web_search_2025_08_26" external_web_access: optional boolean (default: true) Allow live internet access for web search. Defaults to true when omitted. When false, the web search tool runs in offline/cache-only mode and will not fetch new external content.
filters: optional object or null Filters for the search.
allowed_domains: optional array of string or null (default: []) Allowed domains for the search. If not provided, all domains are allowed.
Subdomains of the provided domains are allowed as well.
Example: ["pubmed.ncbi.nlm.nih.gov"]
search_context_size: optional "low" or "medium" or "high" (default: "medium") High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the
United States. To avoid this fallback, pass {"type": "approximate"} without
location fields. To localize results, provide the relevant location fields.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
type: optional "approximate" (default: "approximate") The type of location approximation. Always approximate.
Mcp { server_label, type, allowed_callers, 9 more } Give the model access to additional tools via remote Model Context Protocol
(MCP) servers. Learn more about MCP.
server_label: string A label for this MCP server, used to identify it in tool calls.
type: "mcp" The type of the MCP tool. Always mcp.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" allowed_tools: optional array of string or McpToolFilter or null List of allowed tool names or a filter object.
One of the following: McpAllowedTools = array of string A string array of allowed tool names
McpToolFilter { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
authorization: optional string An OAuth access token that can be used with a remote MCP server, either
with a custom MCP server URL or a service connector. Your application
must handle the OAuth authorization flow and provide the token here.
connector_id: optional "connector_dropbox" or "connector_gmail" or "connector_googlecalendar" or 5 more Deprecated.
Identifier for service connectors, like those available in ChatGPT. One of
server_url, connector_id, or tunnel_id must be provided. Learn more
about service connectors here.
This field is deprecated for models released after September 1, 2026.
Use server_url to connect to a remote MCP server, or tunnel_id to
connect through a Secure MCP Tunnel.
Currently supported connector_id values are:
Dropbox: connector_dropbox
Gmail: connector_gmail
Google Calendar: connector_googlecalendar
Google Drive: connector_googledrive
Microsoft Teams: connector_microsoftteams
Outlook Calendar: connector_outlookcalendar
Outlook Email: connector_outlookemail
SharePoint: connector_sharepoint
One of the following: "connector_dropbox" "connector_gmail" "connector_googlecalendar" "connector_googledrive" "connector_microsoftteams" "connector_outlookcalendar" "connector_outlookemail" "connector_sharepoint" defer_loading: optional boolean Whether this MCP tool is deferred and discovered via tool search.
headers: optional map[string] or null Optional HTTP headers to send to the MCP server. Use for authentication
or other purposes.
require_approval: optional McpToolApprovalFilter or "always" or "never" or null (default: "always") Specify which of the MCP server’s tools require approval.
One of the following: McpToolApprovalFilter { always, never } Specify which of the MCP server’s tools require approval. Can be
always, never, or a filter object associated with tools
that require approval.
always: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
never: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
McpToolApprovalSetting = "always" or "never" Specify a single approval policy for all tools. One of always or
never. When set to always, all tools will require approval. When
set to never, all tools will not require approval.
One of the following: "always" "never" server_description: optional string Optional description of the MCP server, used to provide more context.
server_url: optional string The URL for the MCP server. One of server_url, connector_id, or
tunnel_id must be provided.
formaturi tunnel_id: optional string The Secure MCP Tunnel ID to use instead of a direct server URL. One of
server_url, connector_id, or tunnel_id must be provided.
pattern^tunnel_[a-z0-9]{32}$ CodeInterpreter { container, type, allowed_callers } A tool that runs Python code to help generate a response to a prompt.
container: string or CodeInterpreterToolAuto The code interpreter container. Can be a container ID or an object that
specifies uploaded file IDs to make available to your code, along with an
optional memory_limit setting.
One of the following: string The container ID.
CodeInterpreterToolAuto { type, file_ids, memory_limit, network_policy } Configuration for a code interpreter container. Optionally specify the IDs of the files to run the code on.
type: "auto" (default: "auto") Always auto.
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the code interpreter container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 type: "code_interpreter" The type of the code interpreter tool. Always code_interpreter.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" ProgrammaticToolCalling { type } type: "programmatic_tool_calling" (default: "programmatic_tool_calling") The type of the tool. Always programmatic_tool_calling.
ImageGeneration { type, action, background, 13 more } A tool that generates images using the GPT image models.
type: "image_generation" The type of the image generation tool. Always image_generation.
action: optional "generate" or "edit" or "auto" Whether to generate a new image or edit an existing image. Default: auto.
One of the following: "generate" "edit" "auto" background: optional "transparent" or "opaque" or "auto" (default: "auto") Set the background of the generated image. One of transparent, opaque,
or auto. gpt-image-2.5-sunburst and gpt-image-2.5-flare, including
their 2026-09-08 snapshots, support opaque and transparent
backgrounds. Transparent backgrounds are available for supported GPT Image
models. For gpt-image-2 and gpt-image-2-2026-04-21, this support is in
preview. When using transparent, set the output format to png or webp.
Default: auto.
One of the following: "transparent" "opaque" "auto" complexity: optional "low" or "medium" or "high" Computation spent planning the image, independently of rendering quality.
Overrides the outer response’s reasoning.effort for image generation.
When omitted, inherits a supported outer reasoning effort or uses the
image model’s configured default.
For SVG output, controls the reference image, not the SVG author’s
reasoning.
One of the following: "low" "medium" "high" frame_count: optional integer (default: 9) Number of GIF frames. Requires output_format to be gif. Default: 9.
minimum2 maximum16 frame_duration_ms: optional integer (default: 200) Duration of each GIF frame in milliseconds, rounded to 10 ms.
Requires output_format to be gif. Default: 200.
minimum20 maximum2000 input_fidelity: optional "high" or "low" or null Control how much effort the model will exert to match the style and features, especially facial features, of input images. Supports high and low on gpt-image-1.5; gpt-image-1-mini supports only low. For gpt-image-2, omit this parameter. Defaults to low on supported models.
One of the following: "high" "low" input_image_mask: optional { file_id, image_url } Optional mask for inpainting. Contains image_url
(string, optional) and file_id (string, optional).
file_id: optional string File ID for the mask image.
image_url: optional string Base64-encoded mask image.
model: optional string or "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: string "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: "gpt-image-1-mini" "gpt-image-2" "gpt-image-2-2026-04-21" "gpt-image-2.5-sunburst" "gpt-image-2.5-sunburst-2026-09-08" "gpt-image-2.5-flare" "gpt-image-2.5-flare-2026-09-08" "gpt-image-2.6-sunburst" "gpt-image-2.6-flare" "gpt-image-1.5" "chatgpt-image-latest" "gpt-image-1" Deprecated. GPT Image 1 was shut down on October 23, 2026. Use gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
moderation: optional "auto" or "low" (default: "auto") Moderation level for the generated image. Default: auto.
One of the following: "auto" "low" output_compression: optional integer (default: 100) Compression level for the output image. Default: 100.
minimum0 maximum100 output_format: optional "png" or "webp" or "jpeg" or 2 more (default: "png") The output format of the generated image. One of png, webp, jpeg,
gif, or svg. Default: png. GIF and SVG output require a model
that supports the selected format. GIF output does not support
partial image streaming (partial_images greater than 0),
input_image_mask, or custom output_compression.
Responses lifecycle events can still be streamed for GIF output.
One of the following: "png" "webp" "jpeg" "gif" "svg" partial_images: optional integer (default: 0) Number of partial images to generate in streaming mode, from 0 (default value) to 3.
minimum0 maximum3 quality: optional "low" or "medium" or "high" or 3 more (default: "auto") The quality of the generated image. The GPT image models support low,
medium, and high. gpt-image-2.5-sunburst and gpt-image-2.5-flare,
including their 2026-09-08 snapshots, also support xhigh and max.
Default: auto.
One of the following: "low" "medium" "high" "xhigh" "max" "auto" size: optional string or "1024x1024" or "1024x1536" or "1536x1024" or "auto" (default: "auto") The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: string "1024x1024" or "1024x1536" or "1536x1024" or "auto" The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: "1024x1024" "1024x1536" "1536x1024" "auto" svg: optional { model, reasoning } SVG author model settings. Requires output_format to be svg.
When SVG model selection is enabled, omitting this object uses the
outer response model. Otherwise, the default SVG author is used and
explicit model overrides are unavailable.
model: optional string minLength1 reasoning: optional object or null Reasoning effort for SVG authoring. When omitted, uses the author model’s default in both Images and Responses.
effort: optional "none" or "minimal" or "low" or 4 more Controls how much effort the model should spend on reasoning for this request.
One of the following: "none" "minimal" "low" "medium" "high" "xhigh" "max" LocalShell { type } A tool that allows the model to execute shell commands in a local environment.
type: "local_shell" (default: "local_shell") The type of the local shell tool. Always local_shell.
Shell { type, allowed_callers, environment } A tool that allows the model to execute shell commands.
type: "shell" (default: "shell") The type of the shell tool. Always shell.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" environment: optional ContainerAuto or LocalEnvironment or ContainerReference or null One of the following: ContainerAuto { type, file_ids, memory_limit, 2 more } type: "container_auto" (default: "container_auto") Automatically creates a container for this request
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 skills: optional array of SkillReference or InlineSkill An optional list of skills referenced by id or inline data.
maxItems200 One of the following: SkillReference { skill_id, type, version } skill_id: string The ID of the referenced skill.
minLength1 maxLength64 type: "skill_reference" (default: "skill_reference") References a skill created with the /v1/skills endpoint.
version: optional string Optional skill version. Use a positive integer or ‘latest’. Omit for default.
InlineSkill { description, name, source, type } description: string The description of the skill.
name: string The name of the skill.
source: InlineSkillSource { data, media_type, type } Inline skill payload
type: "inline" (default: "inline") Defines an inline skill for this request.
LocalEnvironment { type, skills } type: "local" (default: "local") Use a local computer environment.
skills: optional array of LocalSkill { description, name, path } An optional list of skills.
maxItems200 description: string The description of the skill.
name: string The name of the skill.
path: string The path to the directory containing the skill.
ContainerReference { container_id, type } container_id: string The ID of the referenced container.
type: "container_reference" (default: "container_reference") References a container created with the /v1/containers endpoint
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
ToolSearchOutputNamespaceTool { description, name, tools, type } Groups function/custom tools under a shared namespace.
description: string A description of the namespace shown to the model.
name: string The namespace name used in tool calls (for example, crm).
minLength1 tools: array of Function or Custom The function/custom tools loaded inside this namespace.
minItems1 One of the following: Function { name, type, allowed_callers, 6 more } name: string The name of the loaded function tool.
pattern^(?![\s\S]*[\r\n])[a-zA-Z0-9_.-]+$ minLength1 maxLength128 type: "function" (default: "function") allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this function should be deferred and discovered via tool search.
description: optional string or null output_schema: optional map[unknown] or null A JSON Schema describing the JSON value encoded in string outputs for this function tool. This does not describe content-array outputs.
parameters: optional unknown or null strict: optional boolean or null Whether to enforce strict parameter validation. If omitted, Responses attempts to use strict validation when the schema is compatible, and falls back to non-strict validation otherwise.
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
type: "namespace" (default: "namespace") The type of the tool. Always namespace.
ToolSearch { type, description, execution, parameters } Hosted or BYOT tool search configuration for deferred tools.
type: "tool_search" (default: "tool_search") The type of the tool. Always tool_search.
description: optional string or null Description shown to the model for a client-executed tool search tool.
execution: optional "server" or "client" Whether tool search is executed by the server or by the client.
One of the following: "server" "client" parameters: optional unknown or null Parameter schema for a client-executed tool search tool.
WebSearchPreview { type, search_content_types, search_context_size, user_location } This tool searches the web for relevant results to use in a response. Learn more about the web search tool.
type: "web_search_preview" or "web_search_preview_2025_03_11" (default: "web_search_preview") The type of the web search tool. One of web_search_preview or web_search_preview_2025_03_11.
One of the following: "web_search_preview" "web_search_preview_2025_03_11" search_content_types: optional array of "text" or "image" One of the following: "text" "image" search_context_size: optional "low" or "medium" or "high" High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the United States. To avoid this fallback, pass {"type": "approximate"} without location fields. To localize results, provide the relevant location fields.
type: "approximate" (default: "approximate") The type of location approximation. Always approximate.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
ApplyPatch { type, allowed_callers } Allows the assistant to create, delete, or update files using unified diffs.
type: "apply_patch" (default: "apply_patch") The type of the tool. Always apply_patch.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" type: "tool_search_output" (default: "tool_search_output") The item type. Always tool_search_output.
id: optional string or null The unique ID of this tool search output.
call_id: optional string or null The unique ID of the tool search call generated by the model.
minLength1 maxLength64 execution: optional "server" or "client" Whether tool search was executed by the server or by the client.
One of the following: "server" "client" status: optional "in_progress" or "completed" or "incomplete" or null The status of the tool search output.
One of the following: "in_progress" "completed" "incomplete" AdditionalTools { role, tools, type, id } role: "developer" (default: "developer") The role that provided the additional tools. Only developer is supported.
tools: array of Function or FileSearch or Computer or 13 more A list of additional tools made available at this item.
One of the following: Function { name, parameters, strict, 6 more } Defines a function in your own code the model can choose to call. Learn more about function calling.
name: string The name of the function to call.
parameters: map[unknown] or null A JSON schema object describing the parameters of the function.
strict: boolean or null Whether strict parameter validation is enforced for this function tool.
type: "function" (default: "function") The type of the function tool. Always function.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
One of the following: "direct" "programmatic" async: optional boolean defer_loading: optional boolean Whether this function is deferred and loaded via tool search.
description: optional string or null A description of the function. Used by the model to determine whether or not to call the function.
output_schema: optional map[unknown] or null A JSON schema object describing the JSON value encoded in string outputs for this function.
FileSearch { type, vector_store_ids, filters, 2 more } A tool that searches for relevant content from uploaded files. Learn more about the file search tool.
type: "file_search" (default: "file_search") The type of the file search tool. Always file_search.
vector_store_ids: array of string The IDs of the vector stores to search.
filters: optional ComparisonFilter or CompoundFilter or null A filter to apply.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } Combine multiple filters using and or or.
filters: array of ComparisonFilter or CompoundFilter Array of filters to combine. Items can be ComparisonFilter or CompoundFilter.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } This type repeats the enclosing schema. type: "and" or "or" Type of operation: and or or.
One of the following: "and" "or" max_num_results: optional integer The maximum number of results to return. This number should be between 1 and 50 inclusive.
ranking_options: optional { hybrid_search, ranker, score_threshold } Ranking options for search.
hybrid_search: optional { embedding_weight, text_weight } Weights that control how reciprocal rank fusion balances semantic embedding matches versus sparse keyword matches when hybrid search is enabled.
embedding_weight: number The weight of the embedding in the reciprocal ranking fusion.
text_weight: number The weight of the text in the reciprocal ranking fusion.
ranker: optional "auto" or "default-2024-11-15" The ranker to use for the file search.
One of the following: "auto" "default-2024-11-15" score_threshold: optional number The score threshold for the file search, a number between 0 and 1. Numbers closer to 1 will attempt to return only the most relevant results, but may return fewer results.
Computer { type } A tool that controls a virtual computer. Learn more about the computer tool.
type: "computer" (default: "computer") The type of the computer tool. Always computer.
ComputerUsePreview { display_height, display_width, environment, type } A tool that controls a virtual computer. Learn more about the computer tool.
display_height: integer The height of the computer display.
display_width: integer The width of the computer display.
environment: "windows" or "mac" or "linux" or 2 more The type of computer environment to control.
One of the following: "windows" "mac" "linux" "ubuntu" "browser" type: "computer_use_preview" (default: "computer_use_preview") The type of the computer use tool. Always computer_use_preview.
WebSearch { type, external_web_access, filters, 2 more } Search the Internet for sources related to the prompt. Learn more about the
web search tool.
type: "web_search" or "web_search_2025_08_26" (default: "web_search") The type of the web search tool. One of web_search or web_search_2025_08_26.
One of the following: "web_search" "web_search_2025_08_26" external_web_access: optional boolean (default: true) Allow live internet access for web search. Defaults to true when omitted. When false, the web search tool runs in offline/cache-only mode and will not fetch new external content.
filters: optional object or null Filters for the search.
allowed_domains: optional array of string or null (default: []) Allowed domains for the search. If not provided, all domains are allowed.
Subdomains of the provided domains are allowed as well.
Example: ["pubmed.ncbi.nlm.nih.gov"]
search_context_size: optional "low" or "medium" or "high" (default: "medium") High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the
United States. To avoid this fallback, pass {"type": "approximate"} without
location fields. To localize results, provide the relevant location fields.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
type: optional "approximate" (default: "approximate") The type of location approximation. Always approximate.
Mcp { server_label, type, allowed_callers, 9 more } Give the model access to additional tools via remote Model Context Protocol
(MCP) servers. Learn more about MCP.
server_label: string A label for this MCP server, used to identify it in tool calls.
type: "mcp" The type of the MCP tool. Always mcp.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" allowed_tools: optional array of string or McpToolFilter or null List of allowed tool names or a filter object.
One of the following: McpAllowedTools = array of string A string array of allowed tool names
McpToolFilter { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
authorization: optional string An OAuth access token that can be used with a remote MCP server, either
with a custom MCP server URL or a service connector. Your application
must handle the OAuth authorization flow and provide the token here.
connector_id: optional "connector_dropbox" or "connector_gmail" or "connector_googlecalendar" or 5 more Deprecated.
Identifier for service connectors, like those available in ChatGPT. One of
server_url, connector_id, or tunnel_id must be provided. Learn more
about service connectors here.
This field is deprecated for models released after September 1, 2026.
Use server_url to connect to a remote MCP server, or tunnel_id to
connect through a Secure MCP Tunnel.
Currently supported connector_id values are:
Dropbox: connector_dropbox
Gmail: connector_gmail
Google Calendar: connector_googlecalendar
Google Drive: connector_googledrive
Microsoft Teams: connector_microsoftteams
Outlook Calendar: connector_outlookcalendar
Outlook Email: connector_outlookemail
SharePoint: connector_sharepoint
One of the following: "connector_dropbox" "connector_gmail" "connector_googlecalendar" "connector_googledrive" "connector_microsoftteams" "connector_outlookcalendar" "connector_outlookemail" "connector_sharepoint" defer_loading: optional boolean Whether this MCP tool is deferred and discovered via tool search.
headers: optional map[string] or null Optional HTTP headers to send to the MCP server. Use for authentication
or other purposes.
require_approval: optional McpToolApprovalFilter or "always" or "never" or null (default: "always") Specify which of the MCP server’s tools require approval.
One of the following: McpToolApprovalFilter { always, never } Specify which of the MCP server’s tools require approval. Can be
always, never, or a filter object associated with tools
that require approval.
always: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
never: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
McpToolApprovalSetting = "always" or "never" Specify a single approval policy for all tools. One of always or
never. When set to always, all tools will require approval. When
set to never, all tools will not require approval.
One of the following: "always" "never" server_description: optional string Optional description of the MCP server, used to provide more context.
server_url: optional string The URL for the MCP server. One of server_url, connector_id, or
tunnel_id must be provided.
formaturi tunnel_id: optional string The Secure MCP Tunnel ID to use instead of a direct server URL. One of
server_url, connector_id, or tunnel_id must be provided.
pattern^tunnel_[a-z0-9]{32}$ CodeInterpreter { container, type, allowed_callers } A tool that runs Python code to help generate a response to a prompt.
container: string or CodeInterpreterToolAuto The code interpreter container. Can be a container ID or an object that
specifies uploaded file IDs to make available to your code, along with an
optional memory_limit setting.
One of the following: string The container ID.
CodeInterpreterToolAuto { type, file_ids, memory_limit, network_policy } Configuration for a code interpreter container. Optionally specify the IDs of the files to run the code on.
type: "auto" (default: "auto") Always auto.
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the code interpreter container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 type: "code_interpreter" The type of the code interpreter tool. Always code_interpreter.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" ProgrammaticToolCalling { type } type: "programmatic_tool_calling" (default: "programmatic_tool_calling") The type of the tool. Always programmatic_tool_calling.
ImageGeneration { type, action, background, 13 more } A tool that generates images using the GPT image models.
type: "image_generation" The type of the image generation tool. Always image_generation.
action: optional "generate" or "edit" or "auto" Whether to generate a new image or edit an existing image. Default: auto.
One of the following: "generate" "edit" "auto" background: optional "transparent" or "opaque" or "auto" (default: "auto") Set the background of the generated image. One of transparent, opaque,
or auto. gpt-image-2.5-sunburst and gpt-image-2.5-flare, including
their 2026-09-08 snapshots, support opaque and transparent
backgrounds. Transparent backgrounds are available for supported GPT Image
models. For gpt-image-2 and gpt-image-2-2026-04-21, this support is in
preview. When using transparent, set the output format to png or webp.
Default: auto.
One of the following: "transparent" "opaque" "auto" complexity: optional "low" or "medium" or "high" Computation spent planning the image, independently of rendering quality.
Overrides the outer response’s reasoning.effort for image generation.
When omitted, inherits a supported outer reasoning effort or uses the
image model’s configured default.
For SVG output, controls the reference image, not the SVG author’s
reasoning.
One of the following: "low" "medium" "high" frame_count: optional integer (default: 9) Number of GIF frames. Requires output_format to be gif. Default: 9.
minimum2 maximum16 frame_duration_ms: optional integer (default: 200) Duration of each GIF frame in milliseconds, rounded to 10 ms.
Requires output_format to be gif. Default: 200.
minimum20 maximum2000 input_fidelity: optional "high" or "low" or null Control how much effort the model will exert to match the style and features, especially facial features, of input images. Supports high and low on gpt-image-1.5; gpt-image-1-mini supports only low. For gpt-image-2, omit this parameter. Defaults to low on supported models.
One of the following: "high" "low" input_image_mask: optional { file_id, image_url } Optional mask for inpainting. Contains image_url
(string, optional) and file_id (string, optional).
file_id: optional string File ID for the mask image.
image_url: optional string Base64-encoded mask image.
model: optional string or "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: string "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: "gpt-image-1-mini" "gpt-image-2" "gpt-image-2-2026-04-21" "gpt-image-2.5-sunburst" "gpt-image-2.5-sunburst-2026-09-08" "gpt-image-2.5-flare" "gpt-image-2.5-flare-2026-09-08" "gpt-image-2.6-sunburst" "gpt-image-2.6-flare" "gpt-image-1.5" "chatgpt-image-latest" "gpt-image-1" Deprecated. GPT Image 1 was shut down on October 23, 2026. Use gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
moderation: optional "auto" or "low" (default: "auto") Moderation level for the generated image. Default: auto.
One of the following: "auto" "low" output_compression: optional integer (default: 100) Compression level for the output image. Default: 100.
minimum0 maximum100 output_format: optional "png" or "webp" or "jpeg" or 2 more (default: "png") The output format of the generated image. One of png, webp, jpeg,
gif, or svg. Default: png. GIF and SVG output require a model
that supports the selected format. GIF output does not support
partial image streaming (partial_images greater than 0),
input_image_mask, or custom output_compression.
Responses lifecycle events can still be streamed for GIF output.
One of the following: "png" "webp" "jpeg" "gif" "svg" partial_images: optional integer (default: 0) Number of partial images to generate in streaming mode, from 0 (default value) to 3.
minimum0 maximum3 quality: optional "low" or "medium" or "high" or 3 more (default: "auto") The quality of the generated image. The GPT image models support low,
medium, and high. gpt-image-2.5-sunburst and gpt-image-2.5-flare,
including their 2026-09-08 snapshots, also support xhigh and max.
Default: auto.
One of the following: "low" "medium" "high" "xhigh" "max" "auto" size: optional string or "1024x1024" or "1024x1536" or "1536x1024" or "auto" (default: "auto") The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: string "1024x1024" or "1024x1536" or "1536x1024" or "auto" The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: "1024x1024" "1024x1536" "1536x1024" "auto" svg: optional { model, reasoning } SVG author model settings. Requires output_format to be svg.
When SVG model selection is enabled, omitting this object uses the
outer response model. Otherwise, the default SVG author is used and
explicit model overrides are unavailable.
model: optional string minLength1 reasoning: optional object or null Reasoning effort for SVG authoring. When omitted, uses the author model’s default in both Images and Responses.
effort: optional "none" or "minimal" or "low" or 4 more Controls how much effort the model should spend on reasoning for this request.
One of the following: "none" "minimal" "low" "medium" "high" "xhigh" "max" LocalShell { type } A tool that allows the model to execute shell commands in a local environment.
type: "local_shell" (default: "local_shell") The type of the local shell tool. Always local_shell.
Shell { type, allowed_callers, environment } A tool that allows the model to execute shell commands.
type: "shell" (default: "shell") The type of the shell tool. Always shell.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" environment: optional ContainerAuto or LocalEnvironment or ContainerReference or null One of the following: ContainerAuto { type, file_ids, memory_limit, 2 more } type: "container_auto" (default: "container_auto") Automatically creates a container for this request
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 skills: optional array of SkillReference or InlineSkill An optional list of skills referenced by id or inline data.
maxItems200 One of the following: SkillReference { skill_id, type, version } skill_id: string The ID of the referenced skill.
minLength1 maxLength64 type: "skill_reference" (default: "skill_reference") References a skill created with the /v1/skills endpoint.
version: optional string Optional skill version. Use a positive integer or ‘latest’. Omit for default.
InlineSkill { description, name, source, type } description: string The description of the skill.
name: string The name of the skill.
source: InlineSkillSource { data, media_type, type } Inline skill payload
type: "inline" (default: "inline") Defines an inline skill for this request.
LocalEnvironment { type, skills } type: "local" (default: "local") Use a local computer environment.
skills: optional array of LocalSkill { description, name, path } An optional list of skills.
maxItems200 description: string The description of the skill.
name: string The name of the skill.
path: string The path to the directory containing the skill.
ContainerReference { container_id, type } container_id: string The ID of the referenced container.
type: "container_reference" (default: "container_reference") References a container created with the /v1/containers endpoint
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
Namespace { description, name, tools, type } Groups function/custom tools under a shared namespace.
description: string A description of the namespace shown to the model.
name: string The namespace name used in tool calls (for example, crm).
minLength1 tools: array of Function or Custom The function/custom tools available inside this namespace.
minItems1 One of the following: Function { name, type, allowed_callers, 6 more } name: string pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 type: "function" (default: "function") allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this function should be deferred and discovered via tool search.
description: optional string or null output_schema: optional map[unknown] or null A JSON Schema describing the JSON value encoded in string outputs for this function tool. This does not describe content-array outputs.
parameters: optional unknown or null strict: optional boolean or null Whether to enforce strict parameter validation. If omitted, Responses attempts to use strict validation when the schema is compatible, and falls back to non-strict validation otherwise.
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
type: "namespace" (default: "namespace") The type of the tool. Always namespace.
ToolSearch { type, description, execution, parameters } Hosted or BYOT tool search configuration for deferred tools.
type: "tool_search" (default: "tool_search") The type of the tool. Always tool_search.
description: optional string or null Description shown to the model for a client-executed tool search tool.
execution: optional "server" or "client" Whether tool search is executed by the server or by the client.
One of the following: "server" "client" parameters: optional unknown or null Parameter schema for a client-executed tool search tool.
WebSearchPreview { type, search_content_types, search_context_size, user_location } This tool searches the web for relevant results to use in a response. Learn more about the web search tool.
type: "web_search_preview" or "web_search_preview_2025_03_11" (default: "web_search_preview") The type of the web search tool. One of web_search_preview or web_search_preview_2025_03_11.
One of the following: "web_search_preview" "web_search_preview_2025_03_11" search_content_types: optional array of "text" or "image" One of the following: "text" "image" search_context_size: optional "low" or "medium" or "high" High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the United States. To avoid this fallback, pass {"type": "approximate"} without location fields. To localize results, provide the relevant location fields.
type: "approximate" (default: "approximate") The type of location approximation. Always approximate.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
ApplyPatch { type, allowed_callers } Allows the assistant to create, delete, or update files using unified diffs.
type: "apply_patch" (default: "apply_patch") The type of the tool. Always apply_patch.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" type: "additional_tools" (default: "additional_tools") The item type. Always additional_tools.
id: optional string or null The unique ID of this additional tools item.
ConfigurationUpdate { type, id, reasoning, text } An update to the conversation’s response configuration. Supply reasoning,
text, or both. Each setting stays in effect until a later update replaces it.
type: "configuration_update" (default: "configuration_update") The item type. Always configuration_update.
id: optional string or null The unique ID of the configuration update item.
reasoning: optional { effort } Updates to reasoning configuration. Only effort is supported.
effort: optional ReasoningEffort or null (default: "medium") The reasoning effort to use for subsequent responses until another
configuration update replaces it.
text: optional { format } The text output configuration for subsequent responses.
format: ResponseFormatText or ResponseFormatTextJSONSchemaConfig Use a JSON schema for structured output or text to return to plain text.
One of the following: ResponseFormatText { type } Default response format. Used to generate text responses.
type: "text" The type of response format being defined. Always text.
ResponseFormatTextJSONSchemaConfig { name, schema, type, 2 more } JSON Schema response format. Used to generate structured JSON responses.
Learn more about Structured Outputs.
name: string The name of the response format. Must be a-z, A-Z, 0-9, or contain
underscores and dashes, with a maximum length of 64.
schema: map[unknown] The schema for the response format, described as a JSON Schema object.
Learn how to build JSON schemas here.
type: "json_schema" The type of response format being defined. Always json_schema.
description: optional string A description of what the response format is for, used by the model to
determine how to respond in the format.
strict: optional boolean or null (default: false) Whether to enable strict schema adherence when generating the output.
If set to true, the model will always follow the exact schema defined
in the schema field. Only a subset of JSON Schema is supported when
strict is true. To learn more, read the Structured Outputs
guide.
Reasoning { id, summary, type, 3 more } A description of the chain of thought used by a reasoning model while generating
a response. Be sure to include these items in your input to the Responses API
for subsequent turns of a conversation if you are manually
managing context.
id: string The unique identifier of the reasoning content.
summary: array of SummaryTextContent { text, type } Reasoning summary content.
text: string A summary of the reasoning output from the model so far.
type: "summary_text" (default: "summary_text") The type of the object. Always summary_text.
type: "reasoning" The type of the object. Always reasoning.
content: optional array of { text, type } Reasoning text content.
text: string The reasoning text from the model.
type: "reasoning_text" (default: "reasoning_text") The type of the reasoning text. Always reasoning_text.
encrypted_content: optional string or null The encrypted content of the reasoning item. This is populated by default
for reasoning items returned by POST /v1/responses and WebSocket
response.create requests.
When streaming, use the completed reasoning item and its
encrypted_content from the response.output_item.done event in
subsequent requests. The encrypted_content in
response.output_item.added may be incomplete. This is especially
important when store is false or when using Zero Data Retention.
status: optional "in_progress" or "completed" or "incomplete" The status of the item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" Compaction { encrypted_content, type, id } A compaction item generated by the v1/responses/compact API.
encrypted_content: string The encrypted content of the compaction summary.
maxLength104857600 type: "compaction" (default: "compaction") The type of the item. Always compaction.
id: optional string or null The ID of the compaction item.
ImageGenerationCall { id, result, status, 9 more } An image generation request made by the model.
id: string The unique ID of the image generation call.
result: string or null The generated image encoded in base64.
status: "in_progress" or "completed" or "generating" or "failed" The status of the image generation call.
One of the following: "in_progress" "completed" "generating" "failed" type: "image_generation_call" The type of the image generation call. Always image_generation_call.
action: optional "generate" or "edit" or "auto" or null The action used for image generation.
One of the following: "generate" "edit" "auto" background: optional "transparent" or "opaque" or "auto" or null The background setting used for generation.
One of the following: "transparent" "opaque" "auto" error: optional { code, message, param, 2 more } Details of a moderation failure, when available.
code: string or null The error code.
message: string A description of the error.
param: string or null The related request parameter.
type: string The error type.
moderation_details: optional { categories, moderation_stage } Public details for a moderation_blocked error, when available.
categories: array of string Public moderation categories. New category values may be added over time.
moderation_stage: "input" or "output" or "unknown" The stage where moderation blocked the image generation call.
One of the following: "input" "output" "unknown" output_format: optional "png" or "webp" or "jpeg" or 2 more or null The output format used for generation.
One of the following: "png" "webp" "jpeg" "gif" "svg" quality: optional "low" or "medium" or "high" or 3 more or null The quality of the image generated by the image generation tool call. One of low, medium, high, xhigh, max, or auto.
One of the following: "low" "medium" "high" "xhigh" "max" "auto" revised_prompt: optional string or null The prompt that was used after any model prompt rewriting.
size: optional string or "1024x1024" or "1024x1536" or "1536x1024" or null The image dimensions as a WIDTHxHEIGHT string, for example 1536x864.
One of the following: string "1024x1024" or "1024x1536" or "1536x1024" The image dimensions as a WIDTHxHEIGHT string, for example 1536x864.
One of the following: "1024x1024" "1024x1536" "1536x1024" svg_usage: optional { input_tokens, model, output_tokens, 2 more } input_tokens: integer model: string output_tokens: integer reasoning_tokens: integer total_tokens: integer CodeInterpreterCall { id, code, container_id, 3 more } A tool call to run code.
id: string The unique ID of the code interpreter tool call.
code: string or null The code to run, or null if not available.
container_id: string The ID of the container used to run the code.
outputs: array of Logs or Image or null The outputs generated by the code interpreter, such as logs or images.
Can be null if no outputs are available.
One of the following: Logs { logs, type } The logs output from the code interpreter.
logs: string The logs output from the code interpreter.
type: "logs" (default: "logs") The type of the output. Always logs.
Image { type, url } The image output from the code interpreter.
type: "image" (default: "image") The type of the output. Always image.
url: string The URL of the image output from the code interpreter.
formaturi status: "in_progress" or "completed" or "incomplete" or 2 more The status of the code interpreter tool call. Valid values are in_progress, completed, incomplete, interpreting, and failed.
One of the following: "in_progress" "completed" "incomplete" "interpreting" "failed" type: "code_interpreter_call" (default: "code_interpreter_call") The type of the code interpreter tool call. Always code_interpreter_call.
LocalShellCall { id, action, call_id, 2 more } A tool call to run a command on the local shell.
id: string The unique ID of the local shell call.
action: { command, env, type, 3 more } Execute a shell command on the server.
command: array of string The command to run.
env: map[string] Environment variables to set for the command.
type: "exec" (default: "exec") The type of the local shell action. Always exec.
timeout_ms: optional integer or null Optional timeout in milliseconds for the command.
user: optional string or null Optional user to run the command as.
working_directory: optional string or null Optional working directory to run the command in.
call_id: string The unique ID of the local shell tool call generated by the model.
status: "in_progress" or "completed" or "incomplete" The status of the local shell call.
One of the following: "in_progress" "completed" "incomplete" type: "local_shell_call" The type of the local shell call. Always local_shell_call.
LocalShellCallOutput { id, output, type, status } The output of a local shell tool call.
id: string The unique ID of the local shell tool call generated by the model.
output: string A JSON string of the output of the local shell tool call.
type: "local_shell_call_output" The type of the local shell tool call output. Always local_shell_call_output.
status: optional "in_progress" or "completed" or "incomplete" or null The status of the item. One of in_progress, completed, or incomplete.
One of the following: "in_progress" "completed" "incomplete" ShellCall { action, call_id, type, 4 more } A tool representing a request to execute one or more shell commands.
action: { commands, max_output_length, timeout_ms } The shell commands and limits that describe how to run the tool call.
commands: array of string Ordered shell commands for the execution environment to run.
max_output_length: optional integer or null Maximum number of UTF-8 characters to capture from combined stdout and stderr output.
timeout_ms: optional integer or null Maximum wall-clock time in milliseconds to allow the shell commands to run.
call_id: string The unique ID of the shell tool call generated by the model.
minLength1 maxLength64 type: "shell_call" (default: "shell_call") The type of the item. Always shell_call.
id: optional string or null The unique ID of the shell tool call. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
environment: optional LocalEnvironment or ContainerReference or null The environment to execute the shell commands in.
One of the following: LocalEnvironment { type, skills } type: "local" (default: "local") Use a local computer environment.
skills: optional array of LocalSkill { description, name, path } An optional list of skills.
maxItems200 description: string The description of the skill.
name: string The name of the skill.
path: string The path to the directory containing the skill.
ContainerReference { container_id, type } container_id: string The ID of the referenced container.
type: "container_reference" (default: "container_reference") References a container created with the /v1/containers endpoint
status: optional "in_progress" or "completed" or "incomplete" or null The status of the shell call. One of in_progress, completed, or incomplete.
One of the following: "in_progress" "completed" "incomplete" ShellCallOutput { call_id, output, type, 4 more } The streamed output items emitted by a shell tool call.
call_id: string The unique ID of the shell tool call generated by the model.
minLength1 maxLength64 output: array of ResponseFunctionShellCallOutputContent { outcome, stderr, stdout } Captured chunks of stdout and stderr output, along with their associated outcomes.
outcome: Timeout or Exit The exit or timeout outcome associated with this shell call.
One of the following: Timeout { type } Indicates that the shell call exceeded its configured time limit.
type: "timeout" (default: "timeout") The outcome type. Always timeout.
Exit { exit_code, type } Indicates that the shell commands finished and returned an exit code.
exit_code: integer The exit code returned by the shell process.
type: "exit" (default: "exit") The outcome type. Always exit.
stderr: string Captured stderr output for the shell call.
maxLength10485760 stdout: string Captured stdout output for the shell call.
maxLength10485760 type: "shell_call_output" (default: "shell_call_output") The type of the item. Always shell_call_output.
id: optional string or null The unique ID of the shell tool call output. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
max_output_length: optional integer or null The maximum number of UTF-8 characters captured for this shell call’s combined output.
status: optional "in_progress" or "completed" or "incomplete" or null The status of the shell call output.
One of the following: "in_progress" "completed" "incomplete" ApplyPatchCall { call_id, operation, status, 3 more } A tool call representing a request to create, delete, or update files using diff patches.
call_id: string The unique ID of the apply patch tool call generated by the model.
minLength1 maxLength64 operation: CreateFile or DeleteFile or UpdateFile The specific create, delete, or update instruction for the apply_patch tool call.
One of the following: CreateFile { diff, path, type } Instruction for creating a new file via the apply_patch tool.
diff: string Unified diff content to apply when creating the file.
maxLength10485760 path: string Path of the file to create relative to the workspace root.
minLength1 type: "create_file" (default: "create_file") The operation type. Always create_file.
DeleteFile { path, type } Instruction for deleting an existing file via the apply_patch tool.
path: string Path of the file to delete relative to the workspace root.
minLength1 type: "delete_file" (default: "delete_file") The operation type. Always delete_file.
UpdateFile { diff, path, type } Instruction for updating an existing file via the apply_patch tool.
diff: string Unified diff content to apply to the existing file.
maxLength10485760 path: string Path of the file to update relative to the workspace root.
minLength1 type: "update_file" (default: "update_file") The operation type. Always update_file.
status: "in_progress" or "completed" The status of the apply patch tool call. One of in_progress or completed.
One of the following: "in_progress" "completed" type: "apply_patch_call" (default: "apply_patch_call") The type of the item. Always apply_patch_call.
id: optional string or null The unique ID of the apply patch tool call. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
ApplyPatchCallOutput { call_id, status, type, 3 more } The streamed output emitted by an apply patch tool call.
call_id: string The unique ID of the apply patch tool call generated by the model.
minLength1 maxLength64 status: "completed" or "failed" The status of the apply patch tool call output. One of completed or failed.
One of the following: "completed" "failed" type: "apply_patch_call_output" (default: "apply_patch_call_output") The type of the item. Always apply_patch_call_output.
id: optional string or null The unique ID of the apply patch tool call output. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
output: optional string or null Optional human-readable log text from the apply patch tool (e.g., patch results or errors).
maxLength10485760 McpListTools { id, server_label, tools, 2 more } A list of tools available on an MCP server.
id: string The unique ID of the list.
server_label: string The label of the MCP server.
tools: array of { input_schema, name, annotations, description } The tools available on the server.
input_schema: unknown The JSON schema describing the tool’s input.
name: string The name of the tool.
annotations: optional unknown or null Additional annotations about the tool.
description: optional string or null The description of the tool.
type: "mcp_list_tools" The type of the item. Always mcp_list_tools.
error: optional string or null Error message if the server could not list tools.
McpApprovalRequest { id, arguments, name, 2 more } A request for human approval of a tool invocation.
id: string The unique ID of the approval request.
arguments: string A JSON string of arguments for the tool.
name: string The name of the tool to run.
server_label: string The label of the MCP server making the request.
type: "mcp_approval_request" The type of the item. Always mcp_approval_request.
McpApprovalResponse { approval_request_id, approve, type, 2 more } A response to an MCP approval request.
approval_request_id: string The ID of the approval request being answered.
approve: boolean Whether the request was approved.
type: "mcp_approval_response" The type of the item. Always mcp_approval_response.
id: optional string or null The unique ID of the approval response
reason: optional string or null Optional reason for the decision.
McpCall { id, arguments, name, 6 more } An invocation of a tool on an MCP server.
id: string The unique ID of the tool call.
arguments: string A JSON string of the arguments passed to the tool.
name: string The name of the tool that was run.
server_label: string The label of the MCP server running the tool.
type: "mcp_call" The type of the item. Always mcp_call.
approval_request_id: optional string or null Unique identifier for the MCP tool call approval request.
Include this value in a subsequent mcp_approval_response input to approve or reject the corresponding tool call.
error: optional McpToolCallError or null The error from the tool call, if any.
output: optional string or null The output from the tool call.
status: optional "in_progress" or "completed" or "incomplete" or 2 more The status of the tool call. One of in_progress, completed, incomplete, calling, or failed.
One of the following: "in_progress" "completed" "incomplete" "calling" "failed" CustomToolCallOutput { call_id, output, type, 2 more } The output of a custom tool call from your code, being sent back to the model.
call_id: string The call ID, used to map this custom tool call output to a custom tool call.
output: string or array of ResponseInputText or ResponseInputImage or ResponseInputFile The output from the custom tool call generated by your code.
Can be a string or an list of output content.
One of the following: StringOutput = string A string of the output of the custom tool call.
OutputContentList = array of ResponseInputText or ResponseInputImage or ResponseInputFile Text, image, or file output of the custom tool call.
One of the following: ResponseInputText { text, type, prompt_cache_breakpoint } A text input to the model.
text: string The text input to the model.
type: "input_text" (default: "input_text") The type of the input item. Always input_text.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputImage { detail, type, file_id, 2 more } An image input to the model. Learn about image inputs.
detail: ImageDetail The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
type: "input_image" (default: "input_image") The type of the input item. Always input_image.
file_id: optional string or null The ID of the file to be sent to the model.
image_url: optional string or null The URL of the image to be sent to the model. A fully qualified URL or base64 encoded image in a data URL.
formaturi prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputFile { type, detail, file_data, 4 more } A file input to the model.
type: "input_file" (default: "input_file") The type of the input item. Always input_file.
detail: optional "auto" or "low" or "high" The detail level of the file to be sent to the model. Use auto to let the system select the detail level; for GPT-5.6 and later models, auto uses high-quality rendering, which may increase input token usage. Use low for lower-cost rendering, or high to render the file at higher quality. Defaults to auto.
One of the following: "auto" "low" "high" file_data: optional string The content of the file to be sent to the model.
file_id: optional string or null The ID of the file to be sent to the model.
file_url: optional string The URL of the file to be sent to the model.
formaturi filename: optional string The name of the file to be sent to the model.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" type: "custom_tool_call_output" The type of the custom tool call output. Always custom_tool_call_output.
id: optional string The unique ID of the custom tool call output in the OpenAI platform.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
CustomToolCall { call_id, input, name, 5 more } A call to a custom tool created by the model.
call_id: string An identifier used to map this custom tool call to a tool call output.
input: string The input for the custom tool call generated by the model.
name: string The name of the custom tool being called.
type: "custom_tool_call" The type of the custom tool call. Always custom_tool_call.
id: optional string The unique ID of the custom tool call in the OpenAI platform.
async: optional boolean Whether the custom tool call runs asynchronously.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
type: "program" (default: "program") namespace: optional string The namespace of the custom tool being called.
CompactionTrigger { type, id } Compacts the current context. Must be the final input item.
type: "compaction_trigger" (default: "compaction_trigger") The type of the item. Always compaction_trigger.
id: optional string or null The unique ID of this compaction trigger.
ItemReference { id, type } An internal identifier for an item to reference.
id: string The ID of the item to reference.
type: optional "item_reference" or null (default: "item_reference") The type of item to reference. Always item_reference.
Program { id, call_id, code, 2 more } id: string The unique ID of this program item.
call_id: string The stable call ID of the program item.
minLength1 maxLength64 code: string The JavaScript source executed by programmatic tool calling.
maxLength10485760 fingerprint: string Opaque program replay fingerprint that must be round-tripped.
maxLength10485760 type: "program" (default: "program") The item type. Always program.
ProgramOutput { id, call_id, result, 2 more } id: string The unique ID of this program output item.
call_id: string The call ID of the program item.
minLength1 maxLength64 result: string The result produced by the program item.
maxLength10485760 status: "completed" or "incomplete" The terminal status of the program output.
One of the following: "completed" "incomplete" type: "program_output" (default: "program_output") The item type. Always program_output.
type: "response.item.create" (default: "response.item.create") The Live client event type. Always response.item.create.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 ResponseCreateEvent { type, event_id } Request a response from the Live session’s Responses backend, or continue a delegated response waiting for tool results. Requires Responses delegation.
type: "response.create" (default: "response.create") The Live client event type. Always response.create.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 SessionCloseEvent { type, event_id } Request that the Live session close. The terminal session.closed event contains the close reason and final usage.
type: "session.close" (default: "session.close") The Live client event type. Always session.close.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512
CommentaryAppendEvent object { content, delegation_id, type, event_id } Provide context the Live model can communicate to the user, optionally for an existing client delegation. Rejected while an audio fixture is active.
CommentaryAppendEvent object { content, delegation_id, type, event_id }
Provide context the Live model can communicate to the user, optionally for an existing client delegation. Rejected while an audio fixture is active.
content: string Speakable context for the Live model, limited to 500 tokens. Use this for a result the model should communicate; use session.thinking.append for silent context.
delegation_id: string or null Required, nullable. Set null for general session context, or use the ID from session.delegation.created for an existing client delegation. Non-null IDs are not accepted with Responses delegation.
minLength1 type: "session.commentary.append" (default: "session.commentary.append") The Live client event type. Always session.commentary.append.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512
CommentaryAppendedEvent object { end_ms, event_id, start_ms, 2 more } Returned when a session.commentary.append command is accepted into the Live session timeline. Acknowledges the added commentary without guaranteeing exact wording or completed audio playback.
CommentaryAppendedEvent object { end_ms, event_id, start_ms, 2 more }
Returned when a session.commentary.append command is accepted into the Live session timeline. Acknowledges the added commentary without guaranteeing exact wording or completed audio playback.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.commentary.appended" (default: "session.commentary.appended") The event type, always session.commentary.appended.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
DataChannelConfig object { allowed_client_events, allowed_server_events } Control which Live events an untrusted WebRTC frontend can send and receive over its data channel. These restrictions do not apply to trusted sideband connections.
DataChannelConfig object { allowed_client_events, allowed_server_events }
Control which Live events an untrusted WebRTC frontend can send and receive over its data channel. These restrictions do not apply to trusted sideband connections.
allowed_client_events: optional "all" or array of string Client event types that the frontend data channel may send. Use ‘all’ to allow every client event; an empty array allows none. Omission preserves the existing allow-all behavior.
One of the following: "all" array of string maxItems256 allowed_server_events: optional "all" or array of ServerEventSelector Server events that may be sent to the frontend data channel. Use ‘all’ to allow every server event; an empty array allows none. Omission preserves the existing allow-all behavior. Responses events use an object with type ‘response.event’ and a response_event selector.
One of the following: "all" array of ServerEventSelector { type, response_event } maxItems256 type: string The outer Live server event type. Use ‘response.event’ for Responses events.
minLength1 maxLength256 response_event: optional string The nested Responses event type. Required when type is ‘response.event’; forbidden for other event types.
minLength1 maxLength256
DecisionEvent object { delegation_id, event, event_id, 4 more } A Decision API event for a Live delegation or transcript trigger. For a trigger, source_event_id identifies the transcript delta that caused this evaluation.
DecisionEvent object { delegation_id, event, event_id, 4 more }
A Decision API event for a Live delegation or transcript trigger. For a trigger, source_event_id identifies the transcript delta that caused this evaluation.
delegation_id: string or null The Live delegation ID, or null for a transcript trigger.
event: { decision, type } decision: Decision { answers, model, usage } The native Decision API response, including its model, ordered answers, and token usage.
type: "decision.completed" (default: "decision.completed") event_id: string The unique ID of this Live server event.
source_event_id: string or null The event_id of the transcript delta that fired this trigger, or null for a delegation.
source_event_type: "session.input_transcript.delta" or "session.output_transcript.delta" or null One of the following: "session.input_transcript.delta" or "session.output_transcript.delta" One of the following: "session.input_transcript.delta" "session.output_transcript.delta" trigger_id: string or null The Live-assigned ID of this trigger firing, or null for a delegation.
type: "decision.event" (default: "decision.event")
DecisionItemCreateEvent object { item, type, event_id } Append an input item for future decisions without starting an evaluation or changing Live voice context.
DecisionItemCreateEvent object { item, type, event_id }
Append an input item for future decisions without starting an evaluation or changing Live voice context.
item: Message or Message or Message or 5 more One of the following: Message { content, role, id, type } content: string or array of InputText or InputImage or InputJSON The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of InputText or InputImage or InputJSON One of the following: InputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") role: "user" (default: "user") The message role. Always user.
id: optional string or null The unique ID of this message item.
type: optional "message" (default: "message") Message { content, role, id, type } content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "developer" (default: "developer") id: optional string or null The unique ID of this message item.
type: optional "message" (default: "message") Message { content, role, id, type } content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "system" (default: "system") id: optional string or null The unique ID of this message item.
type: optional "message" (default: "message") Message { content, role, id, 2 more } content: string or array of OutputText or Refusal The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of OutputText or Refusal One of the following: OutputText { text, type } text: string The text content.
maxLength10485760 type: "output_text" (default: "output_text") Refusal { refusal, type } refusal: string The refusal text.
maxLength10485760 type: "refusal" (default: "refusal") role: "assistant" (default: "assistant") id: optional string or null The unique ID of this message item.
phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer). For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" type: optional "message" (default: "message") FunctionCall { arguments, call_id, name, 4 more } arguments: string The function arguments as a JSON string.
call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 name: string pattern^[a-zA-Z0-9_.-]+$ minLength1 maxLength128 type: "function_call" (default: "function_call") The item type. Always function_call.
id: optional string or null The local Decision item ID.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" FunctionCallOutput { call_id, output, type, 2 more } call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "function_call_output" (default: "function_call_output") The type of the function tool call output. Always function_call_output.
id: optional string or null The local Decision item ID.
status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCall { call_id, input, name, 4 more } call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 input: string The input to the custom tool call.
name: string The name of the custom tool to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 type: "custom_tool_call" (default: "custom_tool_call") The item type. Always custom_tool_call.
id: optional string or null The local Decision item ID.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCallOutput { call_id, output, type, 2 more } call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "custom_tool_call_output" (default: "custom_tool_call_output") id: optional string or null The local Decision item ID.
status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" type: "decision.item.create" (default: "decision.item.create") event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512
DecisionItemCreatedEvent object { event_id, item, type, client_event_id } An input item was appended for future decisions.
DecisionItemCreatedEvent object { event_id, item, type, client_event_id }
An input item was appended for future decisions.
event_id: string The unique ID of the Live server event.
item: Message or Message or Message or 5 more One of the following: Message { id, content, role, type } id: string content: string or array of InputText or InputImage or InputJSON The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of InputText or InputImage or InputJSON One of the following: InputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") role: "user" (default: "user") The message role. Always user.
type: optional "message" (default: "message") Message { id, content, role, type } id: string content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "developer" (default: "developer") type: optional "message" (default: "message") Message { id, content, role, type } id: string content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "system" (default: "system") type: optional "message" (default: "message") Message { id, content, role, 2 more } id: string content: string or array of OutputText or Refusal The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of OutputText or Refusal One of the following: OutputText { text, type } text: string The text content.
maxLength10485760 type: "output_text" (default: "output_text") Refusal { refusal, type } refusal: string The refusal text.
maxLength10485760 type: "refusal" (default: "refusal") role: "assistant" (default: "assistant") phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer). For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" type: optional "message" (default: "message") FunctionCall { id, arguments, call_id, 4 more } id: string arguments: string The function arguments as a JSON string.
call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 name: string pattern^[a-zA-Z0-9_.-]+$ minLength1 maxLength128 type: "function_call" (default: "function_call") The item type. Always function_call.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" FunctionCallOutput { id, call_id, output, 2 more } id: string call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "function_call_output" (default: "function_call_output") The type of the function tool call output. Always function_call_output.
status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCall { id, call_id, input, 4 more } id: string call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 input: string The input to the custom tool call.
name: string The name of the custom tool to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 type: "custom_tool_call" (default: "custom_tool_call") The item type. Always custom_tool_call.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCallOutput { id, call_id, output, 2 more } id: string call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "custom_tool_call_output" (default: "custom_tool_call_output") status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" type: "decision.item.created" (default: "decision.item.created") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
DecisionItemDeleteEvent object { item_id, type, event_id } Delete an input item from future decisions without starting an evaluation or changing Live voice context.
DecisionItemDeleteEvent object { item_id, type, event_id }
Delete an input item from future decisions without starting an evaluation or changing Live voice context.
item_id: string minLength1 type: "decision.item.delete" (default: "decision.item.delete") event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512
DecisionItemDeletedEvent object { event_id, item_id, type, client_event_id } An input item was removed from future decisions.
DecisionItemDeletedEvent object { event_id, item_id, type, client_event_id }
An input item was removed from future decisions.
event_id: string The unique ID of the Live server event.
item_id: string type: "decision.item.deleted" (default: "decision.item.deleted") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
DecisionQuestion = Choice { choices, instructions, type, name } or Predicate { instructions, type, name } or Score { instructions, levels, type, name }
DecisionQuestion = Choice { choices, instructions, type, name } or Predicate { instructions, type, name } or Score { instructions, levels, type, name }
One of the following: Choice { choices, instructions, type, name } choices: array of DecisionChoiceOption { value, description } minItems2 maxItems255 value: string or boolean One of the following: string boolean description: optional string instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "choice" (default: "choice") name: optional string Predicate { instructions, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "predicate" (default: "predicate") name: optional string Score { instructions, levels, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") levels: array of { label, description } minItems2 maxItems10 label: string description: optional string type: "score" (default: "score") name: optional string
DecisionTrigger object { decisions, event_type, type }
DecisionTrigger object { decisions, event_type, type }
decisions: { questions } Questions evaluated using the current transcript and the session’s shared Decision items.
questions: array of DecisionQuestion Native Decision API questions evaluated together, in order.
minItems1 maxItems64 One of the following: Choice { choices, instructions, type, name } choices: array of DecisionChoiceOption { value, description } minItems2 maxItems255 value: string or boolean One of the following: string boolean description: optional string instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "choice" (default: "choice") name: optional string Predicate { instructions, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "predicate" (default: "predicate") name: optional string Score { instructions, levels, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") levels: array of { label, description } minItems2 maxItems10 label: string description: optional string type: "score" (default: "score") name: optional string event_type: "session.input_transcript.delta" or "session.output_transcript.delta" One of the following: "session.input_transcript.delta" "session.output_transcript.delta" type: "decisions" (default: "decisions")
DecisionsDelegationConfig object { model, questions }
DecisionsDelegationConfig object { model, questions }
model: string minLength1 questions: array of DecisionQuestion Native Decision API questions evaluated together, in order.
minItems1 maxItems64 One of the following: Choice { choices, instructions, type, name } choices: array of DecisionChoiceOption { value, description } minItems2 maxItems255 value: string or boolean One of the following: string boolean description: optional string instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "choice" (default: "choice") name: optional string Predicate { instructions, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "predicate" (default: "predicate") name: optional string Score { instructions, levels, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") levels: array of { label, description } minItems2 maxItems10 label: string description: optional string type: "score" (default: "score") name: optional string
DecisionsDelegationUpdateConfig object { model, questions }
DecisionsDelegationUpdateConfig object { model, questions }
model: optional string minLength1 questions: optional array of DecisionQuestion Native Decision API questions evaluated together, in order.
minItems1 maxItems64 One of the following: Choice { choices, instructions, type, name } choices: array of DecisionChoiceOption { value, description } minItems2 maxItems255 value: string or boolean One of the following: string boolean description: optional string instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "choice" (default: "choice") name: optional string Predicate { instructions, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "predicate" (default: "predicate") name: optional string Score { instructions, levels, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") levels: array of { label, description } minItems2 maxItems10 label: string description: optional string type: "score" (default: "score") name: optional string
DelegationCreatedEvent object { delegation, event_id, offset_ms, 2 more } Returned when the Live model delegates work to your application, Responses, or Decisions. Contains delegation metadata and the position on the session timeline where the work was delegated.
DelegationCreatedEvent object { delegation, event_id, offset_ms, 2 more }
Returned when the Live model delegates work to your application, Responses, or Decisions. Contains delegation metadata and the position on the session timeline where the work was delegated.
delegation: { id, target, type, response_id } The delegated work identifier and destination. This object contains metadata, not the task text.
id: string The unique ID of the delegation. Use this as delegation_id when replying to client-owned work or correlating backend events.
target: "client" or "responses" or "decisions" Where the Live model delegated the work: client, responses, or decisions.
One of the following: "client" or "responses" or "decisions" Where the Live model delegated the work: client, responses, or decisions.
One of the following: "client" "responses" "decisions" type: "delegation" (default: "delegation") The object type, always delegation.
response_id: optional string The ID of the Responses API response associated with a Responses delegation. Omitted for client delegations.
event_id: string The unique ID of the Live server event.
offset_ms: integer The position on the Live session timeline where the delegation was created, in milliseconds from the beginning of the session.
type: "session.delegation.created" (default: "session.delegation.created") The event type, always session.delegation.created.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
Error object { code, message, type, 2 more } Details of an error encountered by the Live session, including the affected parameter or client command when available.
Error object { code, message, type, 2 more }
Details of an error encountered by the Live session, including the affected parameter or client command when available.
code: string A machine-readable code identifying the Live error, such as unknown_parameter.
message: string A human-readable explanation of the Live error.
type: string The category of error, such as invalid_request_error for an invalid Live client command.
client_event_id: optional string The event_id of the client command that caused the error, when supplied.
param: optional string The parameter that caused the error, when applicable, such as session.voice.
ErrorEvent object { error, event_id, type, client_event_id } Reports an error in the Live session, such as an invalid client command. Use error.client_event_id, when present, to identify the command that caused the error.
ErrorEvent object { error, event_id, type, client_event_id }
Reports an error in the Live session, such as an invalid client command. Use error.client_event_id, when present, to identify the command that caused the error.
error: Error { code, message, type, 2 more } Details of the Live error and the client command that caused it, when known.
event_id: string The unique ID of the Live server event.
type: "error" (default: "error") The event type, always error.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
ForkSessionConfig object { audio, client, delegation, store } Overrides for a stored session after connecting to the fork WebSocket. An empty object inherits the stored configuration; do not supply a new model. audio.format applies only to the new WebSocket connection. client overrides are only supported for WebRTC forks.
ForkSessionConfig object { audio, client, delegation, store }
Overrides for a stored session after connecting to the fork WebSocket. An empty object inherits the stored configuration; do not supply a new model. audio.format applies only to the new WebSocket connection. client overrides are only supported for WebRTC forks.
audio: optional { format } Audio format for a WebSocket fork. WebRTC forks negotiate their audio format and must omit this field.
format: optional AudioFormat Audio encoding and sample rate for audio sent and received over a Live WebSocket connection. WebRTC and SIP negotiate their media format separately.
client: optional ClientConfig { data_channel } Frontend data-channel permissions for a WebRTC fork. Omitted permissions inherit the stored values. Not supported for WebSocket forks.
delegation: optional { type, acknowledgement_delay_ms, responses } Overrides for the stored session’s Responses backend. Only supported when the stored session already uses Responses delegation; the delegation type cannot change.
type: "responses" (default: "responses") The delegation owner. Always responses for tasks handled by the Responses API.
acknowledgement_delay_ms: optional integer Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations.
minimum0 maximum800 responses: optional ResponsesDelegationUpdateConfig { instructions, max_output_tokens, model, 6 more } Responses backend settings to update. Omitted settings keep their existing values.
store: optional boolean Whether to store the forked session. Omission inherits the stored session’s setting.
ForkSessionStartEvent object { session, type, event_id } Start a Live session after connecting to a stored session’s fork WebSocket. Send an empty session object to use the stored configuration.
ForkSessionStartEvent object { session, type, event_id }
Start a Live session after connecting to a stored session’s fork WebSocket. Send an empty session object to use the stored configuration.
session: ForkSessionConfig { audio, client, delegation, store } Overrides for a stored session after connecting to the fork WebSocket. An empty object inherits the stored configuration; do not supply a new model. audio.format applies only to the new WebSocket connection. client overrides are only supported for WebRTC forks.
type: "session.start" (default: "session.start") The Live client event type. Always session.start.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512
FunctionTool object { name, type, description, 2 more } A function tool available to the Responses backend when the Live model delegates a task.
FunctionTool object { name, type, description, 2 more }
A function tool available to the Responses backend when the Live model delegates a task.
name: string The name the delegated Responses model uses when calling this function.
type: "function" (default: "function") The tool type. Always function.
description: optional string or null What the function does and when the delegated Responses model should call it.
parameters: optional map[unknown] or null A JSON Schema object describing the arguments accepted by the function.
strict: optional boolean or null Whether the delegated Responses model must follow the function’s parameter schema exactly.
InfoEvent object { code, event_id, message, 2 more } An informational notice about the Live session, such as the event permissions applied to a frontend data channel.
InfoEvent object { code, event_id, message, 2 more }
An informational notice about the Live session, such as the event permissions applied to a frontend data channel.
code: string A machine-readable code for the notice, such as data_channel_permissions.
event_id: string The unique ID of the Live server event.
message: string A human-readable explanation of the Live session notice.
type: "info" (default: "info") The event type, always info.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InitialItem = Developer { content, role, id, 2 more } or User { content, role, id, 2 more } or Assistant { content, role, id, 2 more } A developer, user, or assistant message supplied as text history before the Live session starts.
InitialItem = Developer { content, role, id, 2 more } or User { content, role, id, 2 more } or Assistant { content, role, id, 2 more }
A developer, user, or assistant message supplied as text history before the Live session starts.
One of the following: Developer { content, role, id, 2 more } A developer message included in the initial text history of a Live session.
content: array of { text, type } The message content. Supply exactly one text part for the initial Live conversation history.
minItems1 maxItems1 text: string The message text to include in the Live session’s initial conversation history.
type: optional "input_text" (default: "input_text") The text content type. Always input_text.
role: "developer" (default: "developer") The author of this history message. Always developer.
id: optional string or null An optional identifier for the supplied history message. Live uses the message’s role and text to initialize the conversation.
status: optional "incomplete" or "completed" or null The supplied message’s status. Live uses its text as history and does not resume an incomplete message.
One of the following: "incomplete" "completed" type: optional "message" (default: "message") The history item type. Always message.
User { content, role, id, 2 more } A user message included in the initial text history of a Live session.
content: array of { text, type } The message content. Supply exactly one text part for the initial Live conversation history.
minItems1 maxItems1 text: string The message text to include in the Live session’s initial conversation history.
type: optional "input_text" (default: "input_text") The text content type. Always input_text.
role: "user" (default: "user") The author of this history message. Always user.
id: optional string or null An optional identifier for the supplied history message. Live uses the message’s role and text to initialize the conversation.
status: optional "incomplete" or "completed" or null The supplied message’s status. Live uses its text as history and does not resume an incomplete message.
One of the following: "incomplete" "completed" type: optional "message" (default: "message") The history item type. Always message.
Assistant { content, role, id, 2 more } An assistant message included in the initial text history of a Live session.
content: array of Text or OutputText The message content. Supply exactly one text part for the initial Live conversation history.
minItems1 maxItems1 One of the following: Text { text, type } Assistant text supplied as conversation history when starting a Live session.
text: string The message text to include in the Live session’s initial conversation history.
type: optional "text" (default: "text") The text content type. Always text.
OutputText { text, type } Assistant output text supplied as conversation history when starting a Live session.
text: string The message text to include in the Live session’s initial conversation history.
type: "output_text" (default: "output_text") The text content type. Always output_text.
role: "assistant" (default: "assistant") The author of this history message. Always assistant.
id: optional string or null An optional identifier for the supplied history message. Live uses the message’s role and text to initialize the conversation.
status: optional "incomplete" or "completed" or null The supplied message’s status. Live uses its text as history and does not resume an incomplete message.
One of the following: "incomplete" "completed" type: optional "message" (default: "message") The history item type. Always message.
InputAudioAppendEvent object { audio, type, event_id } Send audio to a Live session over its primary WebSocket. WebRTC and SIP sessions send audio over their media transport.
InputAudioAppendEvent object { audio, type, event_id }
Send audio to a Live session over its primary WebSocket. WebRTC and SIP sessions send audio over their media transport.
audio: string Base64-encoded raw audio in the startup-selected format, without a WAV or other container header. Primary WebSocket only; media transports use their audio track. Audio appends have no acknowledgment. Reflected sideband server events reuse this event type and audio key, with no timestamps or event_id; their audio is always mono PCM16LE at 24 kHz.
minLength1 type: "session.input_audio.append" (default: "session.input_audio.append") The Live client event type. Always session.input_audio.append.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512
InputAudioMutedEvent object { event_id, type, client_event_id } Acknowledges a session.input_audio.mute command. An independent input mute sends silence to the model and preserves sideband audio reflection. Fixture playback does not emit this event.
InputAudioMutedEvent object { event_id, type, client_event_id }
Acknowledges a session.input_audio.mute command. An independent input mute sends silence to the model and preserves sideband audio reflection. Fixture playback does not emit this event.
event_id: string The unique ID of the Live server event.
type: "session.input_audio.muted" (default: "session.input_audio.muted") The event type, always session.input_audio.muted.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InputAudioUnmuteEvent object { type, event_id } Resume audio input to a Live model after muting it. The server acknowledges with session.input_audio.unmuted. An active audio fixture still pauses inference until playback finishes.
InputAudioUnmuteEvent object { type, event_id }
Resume audio input to a Live model after muting it. The server acknowledges with session.input_audio.unmuted. An active audio fixture still pauses inference until playback finishes.
InputAudioUnmutedEvent object { event_id, type, client_event_id } Acknowledges a session.input_audio.unmute command, removing the independent input mute. An active audio fixture still pauses inference until playback finishes. Fixture playback does not emit this event.
InputAudioUnmutedEvent object { event_id, type, client_event_id }
Acknowledges a session.input_audio.unmute command, removing the independent input mute. An active audio fixture still pauses inference until playback finishes. Fixture playback does not emit this event.
event_id: string The unique ID of the Live server event.
type: "session.input_audio.unmuted" (default: "session.input_audio.unmuted") The event type, always session.input_audio.unmuted.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InputTranscriptDeltaEvent object { delta, end_ms, event_id, 3 more } A transcript fragment for user input audio in the Live session. Accumulate fragments in delivery order; these events do not define complete turns or include a transcript-done event.
InputTranscriptDeltaEvent object { delta, end_ms, event_id, 3 more }
A transcript fragment for user input audio in the Live session. Accumulate fragments in delivery order; these events do not define complete turns or include a transcript-done event.
delta: string The transcript text fragment for the audio in this time range. Append fragments in delivery order to build the transcript.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.input_transcript.delta" (default: "session.input_transcript.delta") The event type, always session.input_transcript.delta.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InstructionsAppendEvent object { content, delegation_id, type, event_id } Append developer instructions to the Live conversation while it is running, optionally associating them with an existing client delegation. Rejected while an audio fixture is active. These instructions remain in conversation history through compaction and are included in future backend delegations. For external audio, describe content already delivered so the assistant does not repeat it.
InstructionsAppendEvent object { content, delegation_id, type, event_id }
Append developer instructions to the Live conversation while it is running, optionally associating them with an existing client delegation. Rejected while an audio fixture is active. These instructions remain in conversation history through compaction and are included in future backend delegations. For external audio, describe content already delivered so the assistant does not repeat it.
content: string Instruction text to append, limited to 500 tokens. This is a plain string, not an array of content parts.
delegation_id: string or null Required, nullable. Set null for general session context, or use the ID from session.delegation.created for an existing client delegation. Non-null IDs are not accepted with Responses delegation.
minLength1 type: "session.instructions.append" (default: "session.instructions.append") The Live client event type. Always session.instructions.append.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512
InstructionsAppendedEvent object { end_ms, event_id, start_ms, 2 more } Returned when a session.instructions.append command is accepted into the Live session timeline. Acknowledges the appended instructions without guaranteeing that the model has acted on them.
InstructionsAppendedEvent object { end_ms, event_id, start_ms, 2 more }
Returned when a session.instructions.append command is accepted into the Live session timeline. Acknowledges the appended instructions without guaranteeing that the model has acted on them.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.instructions.appended" (default: "session.instructions.appended") The event type, always session.instructions.appended.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
LiveCreateResponse object { session, transport } The created Live session and its transport. WebRTC creation returns 201 Created with an SDP answer. Outbound SIP creation returns 201 Created after initialization, before the callee necessarily answers, without SDP or trunk credentials.
LiveCreateResponse object { session, transport }
The created Live session and its transport. WebRTC creation returns 201 Created with an SDP answer. Outbound SIP creation returns 201 Created after initialization, before the callee necessarily answers, without SDP or trunk credentials.
session: { id } The newly created Live session. Use its ID for session controls and sideband connections.
id: string Opaque session identifier. Preserve the returned value unchanged, including its prefix.
transport: Webrtc or Sip WebRTC transport with an SDP answer, or SIP transport without SDP or trunk credentials. For SIP, attach a sideband using session.id to receive transport.ringing, transport.answered, and transport.failed notifications.
One of the following: Webrtc { sdp, type } WebRTC transport carrying the offer SDP in a creation request or answer SDP in its response.
sdp: string Session Description Protocol message for the WebRTC connection.
minLength1 type: "webrtc" The transport used for the Live session. Always webrtc.
Sip { type } The SIP transport. Trunk credentials and SDP are not returned.
type: "sip" The transport used for the Live session. Always sip.
MediaSessionConfig object { model, audio, client, 6 more } Startup configuration for a Live media session. Follow the Live prompting guide when writing frontend instructions and the backend prompt under delegation.responses.instructions.
MediaSessionConfig object { model, audio, client, 6 more }
Startup configuration for a Live media session. Follow the Live prompting guide when writing frontend instructions and the backend prompt under delegation.responses.instructions.
model: string or "gpt-live-1" The Live model. Required in the session configuration for every transport; do not pass it as a URL query parameter.
One of the following: string minLength1 "gpt-live-1" The Live model. Required in the session configuration for every transport; do not pass it as a URL query parameter.
audio: optional { output } Startup audio configuration. WebRTC and SIP negotiate their audio format on the media transport.
output: optional { voice } Settings for speech generated by the Live model. Choose the voice before starting the session.
voice: optional string or "alloy" or "ash" or "ballad" or 35 more or CustomVoice The voice used for Live speech, as a built-in voice name or a custom voice object containing its ID. Defaults to marin and cannot change after startup.
One of the following: string "alloy" or "ash" or "ballad" or 35 more The voice used for Live speech, as a built-in voice name or a custom voice object containing its ID. Defaults to marin and cannot change after startup.
One of the following: "alloy" "ash" "ballad" "beacon" "bossa" "brise" "calandria" "cedar" "cinder" "compas" "coral" "delta" "destello" "echo" "flitz" "gleam" "halo" "harema" "hornero" "juni" "luciernaga" "marin" "meridian" "nira" "noeul" "nuri" "pearl" "quartz" "ripple" "sage" "shida" "shimmer" "sillage" "stone" "tempo" "verse" "vesper" "willow" CustomVoice { id } id: string minLength1 maxLength128 client: optional ClientConfig { data_channel } Startup-only capabilities for an untrusted frontend attached to a unified WebRTC session. Trusted sideband connections are unaffected.
delegation: optional ClientDelegation or Responses or Decisions or null Who handles tasks delegated by the Live model. Omitted or null selects your application; use responses or decisions to delegate to the corresponding API.
One of the following: ClientDelegation { type, acknowledgement_delay_ms } Delegate tasks to your application. The Live session emits delegation events that your backend handles.
type: "client" (default: "client") The delegation owner. Always client for tasks handled by your application.
acknowledgement_delay_ms: optional integer Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations.
minimum0 maximum800 Responses { responses, type, acknowledgement_delay_ms } Delegate tasks to a Responses model managed by the Live session.
responses: ResponsesDelegationConfig { model, instructions, max_output_tokens, 6 more } Backend model, prompt, and tools used when the Live session delegates a task to Responses.
type: "responses" (default: "responses") The delegation owner. Always responses for tasks handled by the Responses API.
acknowledgement_delay_ms: optional integer Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations.
minimum0 maximum800 Decisions { decisions, type, acknowledgement_delay_ms } decisions: DecisionsDelegationConfig { model, questions } type: "decisions" (default: "decisions") acknowledgement_delay_ms: optional integer Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400. Updates apply to subsequent delegations.
minimum0 maximum800 input: optional array of InitialItem Ordered text-only history supplied before startup. Supports developer, user, and assistant messages with one text part each; at most 128 messages and a configured rendered-token limit (8,192 by default).
maxItems128 One of the following: Developer { content, role, id, 2 more } A developer message included in the initial text history of a Live session.
content: array of { text, type } The message content. Supply exactly one text part for the initial Live conversation history.
minItems1 maxItems1 text: string The message text to include in the Live session’s initial conversation history.
type: optional "input_text" (default: "input_text") The text content type. Always input_text.
role: "developer" (default: "developer") The author of this history message. Always developer.
id: optional string or null An optional identifier for the supplied history message. Live uses the message’s role and text to initialize the conversation.
status: optional "incomplete" or "completed" or null The supplied message’s status. Live uses its text as history and does not resume an incomplete message.
One of the following: "incomplete" "completed" type: optional "message" (default: "message") The history item type. Always message.
User { content, role, id, 2 more } A user message included in the initial text history of a Live session.
content: array of { text, type } The message content. Supply exactly one text part for the initial Live conversation history.
minItems1 maxItems1 text: string The message text to include in the Live session’s initial conversation history.
type: optional "input_text" (default: "input_text") The text content type. Always input_text.
role: "user" (default: "user") The author of this history message. Always user.
id: optional string or null An optional identifier for the supplied history message. Live uses the message’s role and text to initialize the conversation.
status: optional "incomplete" or "completed" or null The supplied message’s status. Live uses its text as history and does not resume an incomplete message.
One of the following: "incomplete" "completed" type: optional "message" (default: "message") The history item type. Always message.
Assistant { content, role, id, 2 more } An assistant message included in the initial text history of a Live session.
content: array of Text or OutputText The message content. Supply exactly one text part for the initial Live conversation history.
minItems1 maxItems1 One of the following: Text { text, type } Assistant text supplied as conversation history when starting a Live session.
text: string The message text to include in the Live session’s initial conversation history.
type: optional "text" (default: "text") The text content type. Always text.
OutputText { text, type } Assistant output text supplied as conversation history when starting a Live session.
text: string The message text to include in the Live session’s initial conversation history.
type: "output_text" (default: "output_text") The text content type. Always output_text.
role: "assistant" (default: "assistant") The author of this history message. Always assistant.
id: optional string or null An optional identifier for the supplied history message. Live uses the message’s role and text to initialize the conversation.
status: optional "incomplete" or "completed" or null The supplied message’s status. Live uses its text as history and does not resume an incomplete message.
One of the following: "incomplete" "completed" type: optional "message" (default: "message") The history item type. Always message.
instructions: optional string or null Frontend instructions for voice, conversation, interruptions, and when to delegate. Start with the Live prompting guide; put business rules and tool workflows in a separate backend prompt. Limited to 16,384 client-supplied tokens. Omitted or blank instructions use server defaults. Immutable after startup.
safety_identifier: optional string or null A stable identifier for your end user, used for safety monitoring and abuse detection.
maxLength128 store: optional boolean Whether to store the session for later forking and recording download. Defaults to false for new sessions.
triggers: optional array of DecisionTrigger { decisions, event_type, type } At most one Decisions trigger per transcript event type. Triggers run independently of delegation. On update, omit triggers to keep the current configuration, send a list to replace it, or send an empty list to disable triggers.
maxItems2 decisions: { questions } Questions evaluated using the current transcript and the session’s shared Decision items.
questions: array of DecisionQuestion Native Decision API questions evaluated together, in order.
minItems1 maxItems64 One of the following: Choice { choices, instructions, type, name } choices: array of DecisionChoiceOption { value, description } minItems2 maxItems255 value: string or boolean One of the following: string boolean description: optional string instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "choice" (default: "choice") name: optional string Predicate { instructions, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "predicate" (default: "predicate") name: optional string Score { instructions, levels, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") levels: array of { label, description } minItems2 maxItems10 label: string description: optional string type: "score" (default: "score") name: optional string event_type: "session.input_transcript.delta" or "session.output_transcript.delta" One of the following: "session.input_transcript.delta" "session.output_transcript.delta" type: "decisions" (default: "decisions")
MediaSessionForkConfig object { client, delegation, store } Optional overrides for a stored Live session. Omitted settings are inherited. The model, voice, frontend instructions, and prior conversation come from the stored session. WebRTC negotiates its audio format; audio.format is only supported on WebSocket forks.
MediaSessionForkConfig object { client, delegation, store }
Optional overrides for a stored Live session. Omitted settings are inherited. The model, voice, frontend instructions, and prior conversation come from the stored session. WebRTC negotiates its audio format; audio.format is only supported on WebSocket forks.
client: optional ClientConfig { data_channel } Startup-only capabilities for an untrusted frontend attached to a unified WebRTC session. Trusted sideband connections are unaffected.
delegation: optional { type, acknowledgement_delay_ms, responses } Update the Responses backend for an existing Live session without changing delegation ownership.
type: "responses" (default: "responses") The delegation owner. Always responses for tasks handled by the Responses API.
acknowledgement_delay_ms: optional integer Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations.
minimum0 maximum800 responses: optional ResponsesDelegationUpdateConfig { instructions, max_output_tokens, model, 6 more } Responses backend settings to update. Omitted settings keep their existing values.
store: optional boolean Whether to store the forked session. Omission inherits the stored session’s setting.
OutputAudioDeltaEvent object { delta, type, end_ms, start_ms } An audio chunk generated by the Live model. Decode and play primary WebSocket chunks in delivery order using the configured session audio format. Sideband connections receive reflected output audio with timestamps.
OutputAudioDeltaEvent object { delta, type, end_ms, start_ms }
An audio chunk generated by the Live model. Decode and play primary WebSocket chunks in delivery order using the configured session audio format. Sideband connections receive reflected output audio with timestamps.
delta: string Base64-encoded raw audio. Primary WebSocket events use the session’s configured format; reflected sideband events use mono PCM16LE at 24 kHz.
type: "session.output_audio.delta" (default: "session.output_audio.delta") The event type, always session.output_audio.delta.
end_ms: optional integer Exclusive session-relative end in milliseconds. Required on reflected sideband events; omitted on the primary WebSocket. Dropped output frames leave gaps between reflected ranges.
start_ms: optional integer Inclusive session-relative start in milliseconds. Required on reflected sideband events; omitted on the primary WebSocket.
OutputAudioFixturePlayEvent object { audio, type, event_id } Play one complete audio fixture through the session output. Send this event over the primary WebSocket or a server sideband. Audio must be raw mono PCM16 little-endian at 24 kHz, regardless of the session audio format, and at most 30 seconds. The entire payload is validated before model speech is interrupted and input is muted. Only one fixture can be active. fixture.playing acknowledges acceptance and fixture.played reports server mixer drain. These notifications are not exact transport-media boundaries. Input arriving during playback is dropped and not replayed.
OutputAudioFixturePlayEvent object { audio, type, event_id }
Play one complete audio fixture through the session output. Send this event over the primary WebSocket or a server sideband. Audio must be raw mono PCM16 little-endian at 24 kHz, regardless of the session audio format, and at most 30 seconds. The entire payload is validated before model speech is interrupted and input is muted. Only one fixture can be active. fixture.playing acknowledges acceptance and fixture.played reports server mixer drain. These notifications are not exact transport-media boundaries. Input arriving during playback is dropped and not replayed.
audio: string Base64-encoded raw mono PCM16 audio, little-endian, at 24 kHz. Do not include a WAV or other container header.
minLength1 maxLength1920000 type: "session.output_audio.fixture.play" (default: "session.output_audio.fixture.play") event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512
OutputAudioFixturePlayedEvent object { event_id, type, client_event_id } The complete audio fixture has drained from the server mixer and normal inference has resumed. Transport buffers may still contain audio; this event does not prove listener-side completion. The optional client_event_id echoes the play command. This event is not emitted for canceled or failed playback.
OutputAudioFixturePlayedEvent object { event_id, type, client_event_id }
The complete audio fixture has drained from the server mixer and normal inference has resumed. Transport buffers may still contain audio; this event does not prove listener-side completion. The optional client_event_id echoes the play command. This event is not emitted for canceled or failed playback.
event_id: string The unique ID of the Live server event.
type: "session.output_audio.fixture.played" (default: "session.output_audio.fixture.played") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
OutputAudioFixturePlayingEvent object { event_id, type, client_event_id } The existing playback path has accepted the validated audio fixture. This acknowledgment does not identify when the listener receives the first sample. The optional client_event_id echoes the play command.
OutputAudioFixturePlayingEvent object { event_id, type, client_event_id }
The existing playback path has accepted the validated audio fixture. This acknowledgment does not identify when the listener receives the first sample. The optional client_event_id echoes the play command.
event_id: string The unique ID of the Live server event.
type: "session.output_audio.fixture.playing" (default: "session.output_audio.fixture.playing") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
OutputTranscriptDeltaEvent object { delta, end_ms, event_id, 3 more } A transcript fragment for assistant output audio in the Live session. Accumulate fragments in delivery order; these events do not define complete turns or include a transcript-done event.
OutputTranscriptDeltaEvent object { delta, end_ms, event_id, 3 more }
A transcript fragment for assistant output audio in the Live session. Accumulate fragments in delivery order; these events do not define complete turns or include a transcript-done event.
delta: string The transcript text fragment for the audio in this time range. Append fragments in delivery order to build the transcript.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.output_transcript.delta" (default: "session.output_transcript.delta") The event type, always session.output_transcript.delta.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
ResponseEvent object { event, event_id, type, 2 more } A streaming Responses API event from a backend delegated to by the Live session. Use the outer delegation_id to associate the nested stream with its Live delegation.
ResponseEvent object { event, event_id, type, 2 more }
A streaming Responses API event from a backend delegated to by the Live session. Use the outer delegation_id to associate the nested stream with its Live delegation.
event: map[unknown] The nested Responses streaming event. Dispatch on its type field. Response lifecycle snapshots omit input and clear instructions, tools, and output to keep messages small; consume granular output events for the generated content.
event_id: string The unique ID of the Live server event.
type: "response.event" (default: "response.event") The event type, always response.event.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
delegation_id: optional string or null The Live delegation associated with the nested Responses event. May be null or omitted when the event cannot be correlated with a delegation.
ResponseItemCreateEvent object { item, type, event_id } Add an input item to the Live session’s Responses backend. Requires Responses delegation; use response.create to request a response.
ResponseItemCreateEvent object { item, type, event_id }
Add an input item to the Live session’s Responses backend. Requires Responses delegation; use response.create to request a response.
item: EasyInputMessage or Message or ResponseOutputMessage or 30 more An input item to append to the Responses backend conversation, such as a user message or a function tool result.
One of the following: EasyInputMessage { content, role, phase, type } A message input to the model with a role indicating instruction following
hierarchy. Instructions given with the developer or system role take
precedence over instructions given with the user role. Messages with the
assistant role are presumed to have been generated by the model in previous
interactions.
content: string or ResponseInputMessageContentList Text, image, or audio input to the model, used to generate a response.
Can also contain previous assistant responses.
One of the following: TextInput = string A text input to the model.
ResponseInputMessageContentList = array of ResponseInputContent A list of one or many input items to the model, containing different content
types.
One of the following: ResponseInputText { text, type, prompt_cache_breakpoint } A text input to the model.
text: string The text input to the model.
type: "input_text" (default: "input_text") The type of the input item. Always input_text.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputImage { detail, type, file_id, 2 more } An image input to the model. Learn about image inputs.
detail: ImageDetail The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
type: "input_image" (default: "input_image") The type of the input item. Always input_image.
file_id: optional string or null The ID of the file to be sent to the model.
image_url: optional string or null The URL of the image to be sent to the model. A fully qualified URL or base64 encoded image in a data URL.
formaturi prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputFile { type, detail, file_data, 4 more } A file input to the model.
type: "input_file" (default: "input_file") The type of the input item. Always input_file.
detail: optional "auto" or "low" or "high" The detail level of the file to be sent to the model. Use auto to let the system select the detail level; for GPT-5.6 and later models, auto uses high-quality rendering, which may increase input token usage. Use low for lower-cost rendering, or high to render the file at higher quality. Defaults to auto.
One of the following: "auto" "low" "high" file_data: optional string The content of the file to be sent to the model.
file_id: optional string or null The ID of the file to be sent to the model.
file_url: optional string The URL of the file to be sent to the model.
formaturi filename: optional string The name of the file to be sent to the model.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" role: "user" or "assistant" or "system" or "developer" The role of the message input. One of user, assistant, system, or
developer.
One of the following: "user" "assistant" "system" "developer" phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer).
For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend
phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" type: optional "message" The type of the message input. Always message.
Message { content, role, status, type } A message input to the model with a role indicating instruction following
hierarchy. Instructions given with the developer or system role take
precedence over instructions given with the user role.
content: ResponseInputMessageContentList { , , } A list of one or many input items to the model, containing different content
types.
role: "user" or "system" or "developer" The role of the message input. One of user, system, or developer.
One of the following: "user" "system" "developer" status: optional "in_progress" or "completed" or "incomplete" The status of item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" type: optional "message" The type of the message input. Always set to message.
ResponseOutputMessage { id, content, role, 3 more } An output message from the model.
id: string The unique ID of the output message.
content: array of ResponseOutputText or ResponseOutputRefusal The content of the output message.
One of the following: ResponseOutputText { annotations, logprobs, text, type } A text output from the model.
annotations: array of FileCitation or URLCitation or ContainerFileCitation or FilePath The annotations of the text output.
One of the following: FileCitation { file_id, filename, index, type } A citation to a file.
file_id: string The ID of the file.
filename: string The filename of the file cited.
index: integer The index in the output text at which to insert the file citation.
type: "file_citation" (default: "file_citation") The type of the file citation. Always file_citation.
URLCitation { end_index, start_index, title, 2 more } A citation for a web resource used to generate a model response.
end_index: integer The index of the last character of the URL citation in the message.
start_index: integer The index of the first character of the URL citation in the message.
title: string The title of the web resource.
type: "url_citation" (default: "url_citation") The type of the URL citation. Always url_citation.
url: string The URL of the web resource.
formaturi ContainerFileCitation { container_id, end_index, file_id, 3 more } A citation for a container file used to generate a model response.
container_id: string The ID of the container file.
end_index: integer The index of the last character of the container file citation in the message.
file_id: string The ID of the file.
filename: string The filename of the container file cited.
start_index: integer The index of the first character of the container file citation in the message.
type: "container_file_citation" (default: "container_file_citation") The type of the container file citation. Always container_file_citation.
FilePath { file_id, index, type } A path to a file.
file_id: string The ID of the file.
index: integer The index of the file in the list of files.
type: "file_path" The type of the file path. Always file_path.
logprobs: array of { token, bytes, logprob, top_logprobs } token: string bytes: array of number logprob: number top_logprobs: array of { token, bytes, logprob } token: string bytes: array of number logprob: number text: string The text output from the model.
type: "output_text" (default: "output_text") The type of the output text. Always output_text.
ResponseOutputRefusal { refusal, type } A refusal from the model.
refusal: string The refusal explanation from the model.
type: "refusal" (default: "refusal") The type of the refusal. Always refusal.
role: "assistant" The role of the output message. Always assistant.
status: "in_progress" or "completed" or "incomplete" The status of the message input. One of in_progress, completed, or
incomplete. Populated when input items are returned via API.
One of the following: "in_progress" "completed" "incomplete" type: "message" The type of the output message. Always message.
phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer).
For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend
phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" FileSearchCall { id, queries, status, 2 more } The results of a file search tool call. See the
file search guide for more information.
id: string The unique ID of the file search tool call.
queries: array of string The queries used to search for files.
status: "in_progress" or "searching" or "completed" or 2 more The status of the file search tool call. One of in_progress,
searching, incomplete or failed,
One of the following: "in_progress" "searching" "completed" "incomplete" "failed" type: "file_search_call" The type of the file search tool call. Always file_search_call.
results: optional array of object or null The results of the file search tool call.
attributes: optional map[string or number or boolean] or null Set of 16 key-value pairs that can be attached to an object. This can be
useful for storing additional information about the object in a structured
format, and querying for objects via API or the dashboard. Keys are strings
with a maximum length of 64 characters. Values are strings with a maximum
length of 512 characters, booleans, or numbers.
One of the following: string maxLength512 number boolean file_id: optional string The unique ID of the file.
filename: optional string The name of the file.
score: optional number The relevance score of the file - a value between 0 and 1.
formatfloat text: optional string The text that was retrieved from the file.
ComputerCall { id, call_id, pending_safety_checks, 4 more } A tool call to a computer use tool. See the
computer use guide for more information.
id: string The unique ID of the computer call.
call_id: string An identifier used when responding to the tool call with output.
pending_safety_checks: array of { id, code, message } The pending safety checks for the computer call.
id: string The ID of the pending safety check.
code: optional string or null The type of the pending safety check.
message: optional string or null Details about the pending safety check.
status: "in_progress" or "completed" or "incomplete" The status of the item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" type: "computer_call" (default: "computer_call") The type of the computer call. Always computer_call.
action: optional ComputerAction A click action.
actions: optional ComputerActionList { Click, DoubleClick, Drag, 6 more } Flattened batched actions for computer_use. Each action includes an
type discriminator and action-specific fields.
ComputerCallOutput { call_id, output, type, 3 more } The output of a computer tool call.
call_id: string The ID of the computer tool call that produced the output.
minLength1 maxLength64 output: ResponseComputerToolCallOutputScreenshot { type, file_id, image_url } A computer screenshot image used with the computer use tool.
type: "computer_call_output" (default: "computer_call_output") The type of the computer tool call output. Always computer_call_output.
id: optional string or null The ID of the computer tool call output.
acknowledged_safety_checks: optional array of object or null The safety checks reported by the API that have been acknowledged by the developer.
id: string The ID of the pending safety check.
code: optional string or null The type of the pending safety check.
message: optional string or null Details about the pending safety check.
status: optional "in_progress" or "completed" or "incomplete" or null The status of the message input. One of in_progress, completed, or incomplete. Populated when input items are returned via API.
One of the following: "in_progress" "completed" "incomplete" WebSearchCall { id, status, type, action } The results of a web search tool call. See the
web search guide for more information.
id: string The unique ID of the web search tool call.
status: "in_progress" or "searching" or "completed" or 2 more The status of the web search tool call.
One of the following: "in_progress" "searching" "completed" "failed" "incomplete" type: "web_search_call" The type of the web search tool call. Always web_search_call.
action: optional Search or OpenPage or FindInPage An object describing the specific action taken in this web search call.
Includes details on how the model used the web (search, open_page, find_in_page).
One of the following: Search { type, queries, query, sources } Action type “search” - Performs a web search query.
type: "search" The action type.
queries: optional array of string The search queries.
query: optional string Deprecated.
The search query.
sources: optional array of { type, url } The sources used in the search.
type: "url" The type of source. Always url.
url: string The URL of the source.
formaturi OpenPage { type, url } Action type “open_page” - Opens a specific URL from search results.
type: "open_page" The action type.
url: optional string or null The URL opened by the model.
formaturi FindInPage { pattern, type, url } Action type “find_in_page”: Searches for a pattern within a loaded page.
pattern: string The pattern or text to search for within the page.
type: "find_in_page" The action type.
url: optional string or null The URL of the page searched for the pattern, when available.
formaturi FunctionCall { arguments, call_id, name, 6 more } A tool call to run a function. See the
function calling guide for more information.
arguments: string A JSON string of the arguments to pass to the function.
call_id: string The unique ID of the function tool call generated by the model.
name: string The name of the function to run.
type: "function_call" The type of the function tool call. Always function_call.
id: optional string The unique ID of the function tool call.
async: optional boolean Whether the function tool call runs asynchronously.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
type: "program" (default: "program") namespace: optional string The namespace of the function to run.
status: optional "in_progress" or "completed" or "incomplete" The status of the item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" FunctionCallOutput { output, type, id, 5 more } The output of a function tool call.
output: string or array of ResponseInputTextContent or ResponseInputImageContent or ResponseInputFileContent Text, image, or file output of the function tool call.
One of the following: string A JSON string of the output of the function tool call.
maxLength10485760 array of ResponseInputTextContent or ResponseInputImageContent or ResponseInputFileContent An array of content outputs (text, image, file) for the function tool call.
One of the following: ResponseInputTextContent { text, type, prompt_cache_breakpoint } A text input to the model.
text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") The type of the input item. Always input_text.
prompt_cache_breakpoint: optional object or null Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputImageContent { type, detail, file_id, 2 more } An image input to the model. Learn about image inputs
type: "input_image" (default: "input_image") The type of the input item. Always input_image.
detail: optional ImageDetail or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
file_id: optional string or null The ID of the file to be sent to the model.
image_url: optional string or null The URL of the image to be sent to the model. A fully qualified URL or base64 encoded image in a data URL.
maxLength20971520 formaturi prompt_cache_breakpoint: optional object or null Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputFileContent { type, detail, file_data, 4 more } A file input to the model.
type: "input_file" (default: "input_file") The type of the input item. Always input_file.
detail: optional "auto" or "low" or "high" The detail level of the file to be sent to the model. Use auto to let the system select the detail level; for GPT-5.6 and later models, auto uses high-quality rendering, which may increase input token usage. Use low for lower-cost rendering, or high to render the file at higher quality. Defaults to auto.
One of the following: "auto" "low" "high" file_data: optional string or null The base64-encoded data of the file to be sent to the model.
maxLength73400320 file_id: optional string or null The ID of the file to be sent to the model.
file_url: optional string or null The URL of the file to be sent to the model.
formaturi filename: optional string or null The name of the file to be sent to the model.
prompt_cache_breakpoint: optional object or null Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" type: "function_call_output" (default: "function_call_output") The type of the function tool call output. Always function_call_output.
id: optional string or null The unique ID of the function tool call output. Populated when this item is returned via API.
call_id: optional string or null The unique ID of the function tool call generated by the model.
minLength1 maxLength64 caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
name: optional string or null The name of the tool that produced the output.
minLength1 maxLength128 namespace: optional string or null The namespace of the tool that produced the output.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 status: optional "in_progress" or "completed" or "incomplete" or null The status of the item. One of in_progress, completed, or incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" ToolSearchCall { arguments, type, id, 3 more } arguments: unknown The arguments supplied to the tool search call.
type: "tool_search_call" (default: "tool_search_call") The item type. Always tool_search_call.
id: optional string or null The unique ID of this tool search call.
call_id: optional string or null The unique ID of the tool search call generated by the model.
minLength1 maxLength64 execution: optional "server" or "client" Whether tool search was executed by the server or by the client.
One of the following: "server" "client" status: optional "in_progress" or "completed" or "incomplete" or null The status of the tool search call.
One of the following: "in_progress" "completed" "incomplete" ToolSearchOutput { tools, type, id, 3 more } tools: array of Function or FileSearch or Computer or 13 more The loaded tool definitions returned by the tool search output.
One of the following: Function { name, parameters, strict, 6 more } Defines a function in your own code the model can choose to call. Learn more about function calling.
name: string The name of the function to call.
parameters: map[unknown] or null A JSON schema object describing the parameters of the function.
strict: boolean or null Whether strict parameter validation is enforced for this function tool.
type: "function" (default: "function") The type of the function tool. Always function.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
One of the following: "direct" "programmatic" async: optional boolean defer_loading: optional boolean Whether this function is deferred and loaded via tool search.
description: optional string or null A description of the function. Used by the model to determine whether or not to call the function.
output_schema: optional map[unknown] or null A JSON schema object describing the JSON value encoded in string outputs for this function.
FileSearch { type, vector_store_ids, filters, 2 more } A tool that searches for relevant content from uploaded files. Learn more about the file search tool.
type: "file_search" (default: "file_search") The type of the file search tool. Always file_search.
vector_store_ids: array of string The IDs of the vector stores to search.
filters: optional ComparisonFilter or CompoundFilter or null A filter to apply.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } Combine multiple filters using and or or.
filters: array of ComparisonFilter or CompoundFilter Array of filters to combine. Items can be ComparisonFilter or CompoundFilter.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } This type repeats the enclosing schema. type: "and" or "or" Type of operation: and or or.
One of the following: "and" "or" max_num_results: optional integer The maximum number of results to return. This number should be between 1 and 50 inclusive.
ranking_options: optional { hybrid_search, ranker, score_threshold } Ranking options for search.
hybrid_search: optional { embedding_weight, text_weight } Weights that control how reciprocal rank fusion balances semantic embedding matches versus sparse keyword matches when hybrid search is enabled.
embedding_weight: number The weight of the embedding in the reciprocal ranking fusion.
text_weight: number The weight of the text in the reciprocal ranking fusion.
ranker: optional "auto" or "default-2024-11-15" The ranker to use for the file search.
One of the following: "auto" "default-2024-11-15" score_threshold: optional number The score threshold for the file search, a number between 0 and 1. Numbers closer to 1 will attempt to return only the most relevant results, but may return fewer results.
Computer { type } A tool that controls a virtual computer. Learn more about the computer tool.
type: "computer" (default: "computer") The type of the computer tool. Always computer.
ComputerUsePreview { display_height, display_width, environment, type } A tool that controls a virtual computer. Learn more about the computer tool.
display_height: integer The height of the computer display.
display_width: integer The width of the computer display.
environment: "windows" or "mac" or "linux" or 2 more The type of computer environment to control.
One of the following: "windows" "mac" "linux" "ubuntu" "browser" type: "computer_use_preview" (default: "computer_use_preview") The type of the computer use tool. Always computer_use_preview.
WebSearch { type, external_web_access, filters, 2 more } Search the Internet for sources related to the prompt. Learn more about the
web search tool.
type: "web_search" or "web_search_2025_08_26" (default: "web_search") The type of the web search tool. One of web_search or web_search_2025_08_26.
One of the following: "web_search" "web_search_2025_08_26" external_web_access: optional boolean (default: true) Allow live internet access for web search. Defaults to true when omitted. When false, the web search tool runs in offline/cache-only mode and will not fetch new external content.
filters: optional object or null Filters for the search.
allowed_domains: optional array of string or null (default: []) Allowed domains for the search. If not provided, all domains are allowed.
Subdomains of the provided domains are allowed as well.
Example: ["pubmed.ncbi.nlm.nih.gov"]
search_context_size: optional "low" or "medium" or "high" (default: "medium") High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the
United States. To avoid this fallback, pass {"type": "approximate"} without
location fields. To localize results, provide the relevant location fields.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
type: optional "approximate" (default: "approximate") The type of location approximation. Always approximate.
Mcp { server_label, type, allowed_callers, 9 more } Give the model access to additional tools via remote Model Context Protocol
(MCP) servers. Learn more about MCP.
server_label: string A label for this MCP server, used to identify it in tool calls.
type: "mcp" The type of the MCP tool. Always mcp.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" allowed_tools: optional array of string or McpToolFilter or null List of allowed tool names or a filter object.
One of the following: McpAllowedTools = array of string A string array of allowed tool names
McpToolFilter { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
authorization: optional string An OAuth access token that can be used with a remote MCP server, either
with a custom MCP server URL or a service connector. Your application
must handle the OAuth authorization flow and provide the token here.
connector_id: optional "connector_dropbox" or "connector_gmail" or "connector_googlecalendar" or 5 more Deprecated.
Identifier for service connectors, like those available in ChatGPT. One of
server_url, connector_id, or tunnel_id must be provided. Learn more
about service connectors here.
This field is deprecated for models released after September 1, 2026.
Use server_url to connect to a remote MCP server, or tunnel_id to
connect through a Secure MCP Tunnel.
Currently supported connector_id values are:
Dropbox: connector_dropbox
Gmail: connector_gmail
Google Calendar: connector_googlecalendar
Google Drive: connector_googledrive
Microsoft Teams: connector_microsoftteams
Outlook Calendar: connector_outlookcalendar
Outlook Email: connector_outlookemail
SharePoint: connector_sharepoint
One of the following: "connector_dropbox" "connector_gmail" "connector_googlecalendar" "connector_googledrive" "connector_microsoftteams" "connector_outlookcalendar" "connector_outlookemail" "connector_sharepoint" defer_loading: optional boolean Whether this MCP tool is deferred and discovered via tool search.
headers: optional map[string] or null Optional HTTP headers to send to the MCP server. Use for authentication
or other purposes.
require_approval: optional McpToolApprovalFilter or "always" or "never" or null (default: "always") Specify which of the MCP server’s tools require approval.
One of the following: McpToolApprovalFilter { always, never } Specify which of the MCP server’s tools require approval. Can be
always, never, or a filter object associated with tools
that require approval.
always: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
never: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
McpToolApprovalSetting = "always" or "never" Specify a single approval policy for all tools. One of always or
never. When set to always, all tools will require approval. When
set to never, all tools will not require approval.
One of the following: "always" "never" server_description: optional string Optional description of the MCP server, used to provide more context.
server_url: optional string The URL for the MCP server. One of server_url, connector_id, or
tunnel_id must be provided.
formaturi tunnel_id: optional string The Secure MCP Tunnel ID to use instead of a direct server URL. One of
server_url, connector_id, or tunnel_id must be provided.
pattern^tunnel_[a-z0-9]{32}$ CodeInterpreter { container, type, allowed_callers } A tool that runs Python code to help generate a response to a prompt.
container: string or CodeInterpreterToolAuto The code interpreter container. Can be a container ID or an object that
specifies uploaded file IDs to make available to your code, along with an
optional memory_limit setting.
One of the following: string The container ID.
CodeInterpreterToolAuto { type, file_ids, memory_limit, network_policy } Configuration for a code interpreter container. Optionally specify the IDs of the files to run the code on.
type: "auto" (default: "auto") Always auto.
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the code interpreter container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 type: "code_interpreter" The type of the code interpreter tool. Always code_interpreter.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" ProgrammaticToolCalling { type } type: "programmatic_tool_calling" (default: "programmatic_tool_calling") The type of the tool. Always programmatic_tool_calling.
ImageGeneration { type, action, background, 13 more } A tool that generates images using the GPT image models.
type: "image_generation" The type of the image generation tool. Always image_generation.
action: optional "generate" or "edit" or "auto" Whether to generate a new image or edit an existing image. Default: auto.
One of the following: "generate" "edit" "auto" background: optional "transparent" or "opaque" or "auto" (default: "auto") Set the background of the generated image. One of transparent, opaque,
or auto. gpt-image-2.5-sunburst and gpt-image-2.5-flare, including
their 2026-09-08 snapshots, support opaque and transparent
backgrounds. Transparent backgrounds are available for supported GPT Image
models. For gpt-image-2 and gpt-image-2-2026-04-21, this support is in
preview. When using transparent, set the output format to png or webp.
Default: auto.
One of the following: "transparent" "opaque" "auto" complexity: optional "low" or "medium" or "high" Computation spent planning the image, independently of rendering quality.
Overrides the outer response’s reasoning.effort for image generation.
When omitted, inherits a supported outer reasoning effort or uses the
image model’s configured default.
For SVG output, controls the reference image, not the SVG author’s
reasoning.
One of the following: "low" "medium" "high" frame_count: optional integer (default: 9) Number of GIF frames. Requires output_format to be gif. Default: 9.
minimum2 maximum16 frame_duration_ms: optional integer (default: 200) Duration of each GIF frame in milliseconds, rounded to 10 ms.
Requires output_format to be gif. Default: 200.
minimum20 maximum2000 input_fidelity: optional "high" or "low" or null Control how much effort the model will exert to match the style and features, especially facial features, of input images. Supports high and low on gpt-image-1.5; gpt-image-1-mini supports only low. For gpt-image-2, omit this parameter. Defaults to low on supported models.
One of the following: "high" "low" input_image_mask: optional { file_id, image_url } Optional mask for inpainting. Contains image_url
(string, optional) and file_id (string, optional).
file_id: optional string File ID for the mask image.
image_url: optional string Base64-encoded mask image.
model: optional string or "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: string "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: "gpt-image-1-mini" "gpt-image-2" "gpt-image-2-2026-04-21" "gpt-image-2.5-sunburst" "gpt-image-2.5-sunburst-2026-09-08" "gpt-image-2.5-flare" "gpt-image-2.5-flare-2026-09-08" "gpt-image-2.6-sunburst" "gpt-image-2.6-flare" "gpt-image-1.5" "chatgpt-image-latest" "gpt-image-1" Deprecated. GPT Image 1 was shut down on October 23, 2026. Use gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
moderation: optional "auto" or "low" (default: "auto") Moderation level for the generated image. Default: auto.
One of the following: "auto" "low" output_compression: optional integer (default: 100) Compression level for the output image. Default: 100.
minimum0 maximum100 output_format: optional "png" or "webp" or "jpeg" or 2 more (default: "png") The output format of the generated image. One of png, webp, jpeg,
gif, or svg. Default: png. GIF and SVG output require a model
that supports the selected format. GIF output does not support
partial image streaming (partial_images greater than 0),
input_image_mask, or custom output_compression.
Responses lifecycle events can still be streamed for GIF output.
One of the following: "png" "webp" "jpeg" "gif" "svg" partial_images: optional integer (default: 0) Number of partial images to generate in streaming mode, from 0 (default value) to 3.
minimum0 maximum3 quality: optional "low" or "medium" or "high" or 3 more (default: "auto") The quality of the generated image. The GPT image models support low,
medium, and high. gpt-image-2.5-sunburst and gpt-image-2.5-flare,
including their 2026-09-08 snapshots, also support xhigh and max.
Default: auto.
One of the following: "low" "medium" "high" "xhigh" "max" "auto" size: optional string or "1024x1024" or "1024x1536" or "1536x1024" or "auto" (default: "auto") The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: string "1024x1024" or "1024x1536" or "1536x1024" or "auto" The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: "1024x1024" "1024x1536" "1536x1024" "auto" svg: optional { model, reasoning } SVG author model settings. Requires output_format to be svg.
When SVG model selection is enabled, omitting this object uses the
outer response model. Otherwise, the default SVG author is used and
explicit model overrides are unavailable.
model: optional string minLength1 reasoning: optional object or null Reasoning effort for SVG authoring. When omitted, uses the author model’s default in both Images and Responses.
effort: optional "none" or "minimal" or "low" or 4 more Controls how much effort the model should spend on reasoning for this request.
One of the following: "none" "minimal" "low" "medium" "high" "xhigh" "max" LocalShell { type } A tool that allows the model to execute shell commands in a local environment.
type: "local_shell" (default: "local_shell") The type of the local shell tool. Always local_shell.
Shell { type, allowed_callers, environment } A tool that allows the model to execute shell commands.
type: "shell" (default: "shell") The type of the shell tool. Always shell.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" environment: optional ContainerAuto or LocalEnvironment or ContainerReference or null One of the following: ContainerAuto { type, file_ids, memory_limit, 2 more } type: "container_auto" (default: "container_auto") Automatically creates a container for this request
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 skills: optional array of SkillReference or InlineSkill An optional list of skills referenced by id or inline data.
maxItems200 One of the following: SkillReference { skill_id, type, version } skill_id: string The ID of the referenced skill.
minLength1 maxLength64 type: "skill_reference" (default: "skill_reference") References a skill created with the /v1/skills endpoint.
version: optional string Optional skill version. Use a positive integer or ‘latest’. Omit for default.
InlineSkill { description, name, source, type } description: string The description of the skill.
name: string The name of the skill.
source: InlineSkillSource { data, media_type, type } Inline skill payload
type: "inline" (default: "inline") Defines an inline skill for this request.
LocalEnvironment { type, skills } type: "local" (default: "local") Use a local computer environment.
skills: optional array of LocalSkill { description, name, path } An optional list of skills.
maxItems200 description: string The description of the skill.
name: string The name of the skill.
path: string The path to the directory containing the skill.
ContainerReference { container_id, type } container_id: string The ID of the referenced container.
type: "container_reference" (default: "container_reference") References a container created with the /v1/containers endpoint
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
ToolSearchOutputNamespaceTool { description, name, tools, type } Groups function/custom tools under a shared namespace.
description: string A description of the namespace shown to the model.
name: string The namespace name used in tool calls (for example, crm).
minLength1 tools: array of Function or Custom The function/custom tools loaded inside this namespace.
minItems1 One of the following: Function { name, type, allowed_callers, 6 more } name: string The name of the loaded function tool.
pattern^(?![\s\S]*[\r\n])[a-zA-Z0-9_.-]+$ minLength1 maxLength128 type: "function" (default: "function") allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this function should be deferred and discovered via tool search.
description: optional string or null output_schema: optional map[unknown] or null A JSON Schema describing the JSON value encoded in string outputs for this function tool. This does not describe content-array outputs.
parameters: optional unknown or null strict: optional boolean or null Whether to enforce strict parameter validation. If omitted, Responses attempts to use strict validation when the schema is compatible, and falls back to non-strict validation otherwise.
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
type: "namespace" (default: "namespace") The type of the tool. Always namespace.
ToolSearch { type, description, execution, parameters } Hosted or BYOT tool search configuration for deferred tools.
type: "tool_search" (default: "tool_search") The type of the tool. Always tool_search.
description: optional string or null Description shown to the model for a client-executed tool search tool.
execution: optional "server" or "client" Whether tool search is executed by the server or by the client.
One of the following: "server" "client" parameters: optional unknown or null Parameter schema for a client-executed tool search tool.
WebSearchPreview { type, search_content_types, search_context_size, user_location } This tool searches the web for relevant results to use in a response. Learn more about the web search tool.
type: "web_search_preview" or "web_search_preview_2025_03_11" (default: "web_search_preview") The type of the web search tool. One of web_search_preview or web_search_preview_2025_03_11.
One of the following: "web_search_preview" "web_search_preview_2025_03_11" search_content_types: optional array of "text" or "image" One of the following: "text" "image" search_context_size: optional "low" or "medium" or "high" High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the United States. To avoid this fallback, pass {"type": "approximate"} without location fields. To localize results, provide the relevant location fields.
type: "approximate" (default: "approximate") The type of location approximation. Always approximate.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
ApplyPatch { type, allowed_callers } Allows the assistant to create, delete, or update files using unified diffs.
type: "apply_patch" (default: "apply_patch") The type of the tool. Always apply_patch.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" type: "tool_search_output" (default: "tool_search_output") The item type. Always tool_search_output.
id: optional string or null The unique ID of this tool search output.
call_id: optional string or null The unique ID of the tool search call generated by the model.
minLength1 maxLength64 execution: optional "server" or "client" Whether tool search was executed by the server or by the client.
One of the following: "server" "client" status: optional "in_progress" or "completed" or "incomplete" or null The status of the tool search output.
One of the following: "in_progress" "completed" "incomplete" AdditionalTools { role, tools, type, id } role: "developer" (default: "developer") The role that provided the additional tools. Only developer is supported.
tools: array of Function or FileSearch or Computer or 13 more A list of additional tools made available at this item.
One of the following: Function { name, parameters, strict, 6 more } Defines a function in your own code the model can choose to call. Learn more about function calling.
name: string The name of the function to call.
parameters: map[unknown] or null A JSON schema object describing the parameters of the function.
strict: boolean or null Whether strict parameter validation is enforced for this function tool.
type: "function" (default: "function") The type of the function tool. Always function.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
One of the following: "direct" "programmatic" async: optional boolean defer_loading: optional boolean Whether this function is deferred and loaded via tool search.
description: optional string or null A description of the function. Used by the model to determine whether or not to call the function.
output_schema: optional map[unknown] or null A JSON schema object describing the JSON value encoded in string outputs for this function.
FileSearch { type, vector_store_ids, filters, 2 more } A tool that searches for relevant content from uploaded files. Learn more about the file search tool.
type: "file_search" (default: "file_search") The type of the file search tool. Always file_search.
vector_store_ids: array of string The IDs of the vector stores to search.
filters: optional ComparisonFilter or CompoundFilter or null A filter to apply.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } Combine multiple filters using and or or.
filters: array of ComparisonFilter or CompoundFilter Array of filters to combine. Items can be ComparisonFilter or CompoundFilter.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } This type repeats the enclosing schema. type: "and" or "or" Type of operation: and or or.
One of the following: "and" "or" max_num_results: optional integer The maximum number of results to return. This number should be between 1 and 50 inclusive.
ranking_options: optional { hybrid_search, ranker, score_threshold } Ranking options for search.
hybrid_search: optional { embedding_weight, text_weight } Weights that control how reciprocal rank fusion balances semantic embedding matches versus sparse keyword matches when hybrid search is enabled.
embedding_weight: number The weight of the embedding in the reciprocal ranking fusion.
text_weight: number The weight of the text in the reciprocal ranking fusion.
ranker: optional "auto" or "default-2024-11-15" The ranker to use for the file search.
One of the following: "auto" "default-2024-11-15" score_threshold: optional number The score threshold for the file search, a number between 0 and 1. Numbers closer to 1 will attempt to return only the most relevant results, but may return fewer results.
Computer { type } A tool that controls a virtual computer. Learn more about the computer tool.
type: "computer" (default: "computer") The type of the computer tool. Always computer.
ComputerUsePreview { display_height, display_width, environment, type } A tool that controls a virtual computer. Learn more about the computer tool.
display_height: integer The height of the computer display.
display_width: integer The width of the computer display.
environment: "windows" or "mac" or "linux" or 2 more The type of computer environment to control.
One of the following: "windows" "mac" "linux" "ubuntu" "browser" type: "computer_use_preview" (default: "computer_use_preview") The type of the computer use tool. Always computer_use_preview.
WebSearch { type, external_web_access, filters, 2 more } Search the Internet for sources related to the prompt. Learn more about the
web search tool.
type: "web_search" or "web_search_2025_08_26" (default: "web_search") The type of the web search tool. One of web_search or web_search_2025_08_26.
One of the following: "web_search" "web_search_2025_08_26" external_web_access: optional boolean (default: true) Allow live internet access for web search. Defaults to true when omitted. When false, the web search tool runs in offline/cache-only mode and will not fetch new external content.
filters: optional object or null Filters for the search.
allowed_domains: optional array of string or null (default: []) Allowed domains for the search. If not provided, all domains are allowed.
Subdomains of the provided domains are allowed as well.
Example: ["pubmed.ncbi.nlm.nih.gov"]
search_context_size: optional "low" or "medium" or "high" (default: "medium") High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the
United States. To avoid this fallback, pass {"type": "approximate"} without
location fields. To localize results, provide the relevant location fields.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
type: optional "approximate" (default: "approximate") The type of location approximation. Always approximate.
Mcp { server_label, type, allowed_callers, 9 more } Give the model access to additional tools via remote Model Context Protocol
(MCP) servers. Learn more about MCP.
server_label: string A label for this MCP server, used to identify it in tool calls.
type: "mcp" The type of the MCP tool. Always mcp.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" allowed_tools: optional array of string or McpToolFilter or null List of allowed tool names or a filter object.
One of the following: McpAllowedTools = array of string A string array of allowed tool names
McpToolFilter { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
authorization: optional string An OAuth access token that can be used with a remote MCP server, either
with a custom MCP server URL or a service connector. Your application
must handle the OAuth authorization flow and provide the token here.
connector_id: optional "connector_dropbox" or "connector_gmail" or "connector_googlecalendar" or 5 more Deprecated.
Identifier for service connectors, like those available in ChatGPT. One of
server_url, connector_id, or tunnel_id must be provided. Learn more
about service connectors here.
This field is deprecated for models released after September 1, 2026.
Use server_url to connect to a remote MCP server, or tunnel_id to
connect through a Secure MCP Tunnel.
Currently supported connector_id values are:
Dropbox: connector_dropbox
Gmail: connector_gmail
Google Calendar: connector_googlecalendar
Google Drive: connector_googledrive
Microsoft Teams: connector_microsoftteams
Outlook Calendar: connector_outlookcalendar
Outlook Email: connector_outlookemail
SharePoint: connector_sharepoint
One of the following: "connector_dropbox" "connector_gmail" "connector_googlecalendar" "connector_googledrive" "connector_microsoftteams" "connector_outlookcalendar" "connector_outlookemail" "connector_sharepoint" defer_loading: optional boolean Whether this MCP tool is deferred and discovered via tool search.
headers: optional map[string] or null Optional HTTP headers to send to the MCP server. Use for authentication
or other purposes.
require_approval: optional McpToolApprovalFilter or "always" or "never" or null (default: "always") Specify which of the MCP server’s tools require approval.
One of the following: McpToolApprovalFilter { always, never } Specify which of the MCP server’s tools require approval. Can be
always, never, or a filter object associated with tools
that require approval.
always: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
never: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
McpToolApprovalSetting = "always" or "never" Specify a single approval policy for all tools. One of always or
never. When set to always, all tools will require approval. When
set to never, all tools will not require approval.
One of the following: "always" "never" server_description: optional string Optional description of the MCP server, used to provide more context.
server_url: optional string The URL for the MCP server. One of server_url, connector_id, or
tunnel_id must be provided.
formaturi tunnel_id: optional string The Secure MCP Tunnel ID to use instead of a direct server URL. One of
server_url, connector_id, or tunnel_id must be provided.
pattern^tunnel_[a-z0-9]{32}$ CodeInterpreter { container, type, allowed_callers } A tool that runs Python code to help generate a response to a prompt.
container: string or CodeInterpreterToolAuto The code interpreter container. Can be a container ID or an object that
specifies uploaded file IDs to make available to your code, along with an
optional memory_limit setting.
One of the following: string The container ID.
CodeInterpreterToolAuto { type, file_ids, memory_limit, network_policy } Configuration for a code interpreter container. Optionally specify the IDs of the files to run the code on.
type: "auto" (default: "auto") Always auto.
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the code interpreter container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 type: "code_interpreter" The type of the code interpreter tool. Always code_interpreter.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" ProgrammaticToolCalling { type } type: "programmatic_tool_calling" (default: "programmatic_tool_calling") The type of the tool. Always programmatic_tool_calling.
ImageGeneration { type, action, background, 13 more } A tool that generates images using the GPT image models.
type: "image_generation" The type of the image generation tool. Always image_generation.
action: optional "generate" or "edit" or "auto" Whether to generate a new image or edit an existing image. Default: auto.
One of the following: "generate" "edit" "auto" background: optional "transparent" or "opaque" or "auto" (default: "auto") Set the background of the generated image. One of transparent, opaque,
or auto. gpt-image-2.5-sunburst and gpt-image-2.5-flare, including
their 2026-09-08 snapshots, support opaque and transparent
backgrounds. Transparent backgrounds are available for supported GPT Image
models. For gpt-image-2 and gpt-image-2-2026-04-21, this support is in
preview. When using transparent, set the output format to png or webp.
Default: auto.
One of the following: "transparent" "opaque" "auto" complexity: optional "low" or "medium" or "high" Computation spent planning the image, independently of rendering quality.
Overrides the outer response’s reasoning.effort for image generation.
When omitted, inherits a supported outer reasoning effort or uses the
image model’s configured default.
For SVG output, controls the reference image, not the SVG author’s
reasoning.
One of the following: "low" "medium" "high" frame_count: optional integer (default: 9) Number of GIF frames. Requires output_format to be gif. Default: 9.
minimum2 maximum16 frame_duration_ms: optional integer (default: 200) Duration of each GIF frame in milliseconds, rounded to 10 ms.
Requires output_format to be gif. Default: 200.
minimum20 maximum2000 input_fidelity: optional "high" or "low" or null Control how much effort the model will exert to match the style and features, especially facial features, of input images. Supports high and low on gpt-image-1.5; gpt-image-1-mini supports only low. For gpt-image-2, omit this parameter. Defaults to low on supported models.
One of the following: "high" "low" input_image_mask: optional { file_id, image_url } Optional mask for inpainting. Contains image_url
(string, optional) and file_id (string, optional).
file_id: optional string File ID for the mask image.
image_url: optional string Base64-encoded mask image.
model: optional string or "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: string "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: "gpt-image-1-mini" "gpt-image-2" "gpt-image-2-2026-04-21" "gpt-image-2.5-sunburst" "gpt-image-2.5-sunburst-2026-09-08" "gpt-image-2.5-flare" "gpt-image-2.5-flare-2026-09-08" "gpt-image-2.6-sunburst" "gpt-image-2.6-flare" "gpt-image-1.5" "chatgpt-image-latest" "gpt-image-1" Deprecated. GPT Image 1 was shut down on October 23, 2026. Use gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
moderation: optional "auto" or "low" (default: "auto") Moderation level for the generated image. Default: auto.
One of the following: "auto" "low" output_compression: optional integer (default: 100) Compression level for the output image. Default: 100.
minimum0 maximum100 output_format: optional "png" or "webp" or "jpeg" or 2 more (default: "png") The output format of the generated image. One of png, webp, jpeg,
gif, or svg. Default: png. GIF and SVG output require a model
that supports the selected format. GIF output does not support
partial image streaming (partial_images greater than 0),
input_image_mask, or custom output_compression.
Responses lifecycle events can still be streamed for GIF output.
One of the following: "png" "webp" "jpeg" "gif" "svg" partial_images: optional integer (default: 0) Number of partial images to generate in streaming mode, from 0 (default value) to 3.
minimum0 maximum3 quality: optional "low" or "medium" or "high" or 3 more (default: "auto") The quality of the generated image. The GPT image models support low,
medium, and high. gpt-image-2.5-sunburst and gpt-image-2.5-flare,
including their 2026-09-08 snapshots, also support xhigh and max.
Default: auto.
One of the following: "low" "medium" "high" "xhigh" "max" "auto" size: optional string or "1024x1024" or "1024x1536" or "1536x1024" or "auto" (default: "auto") The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: string "1024x1024" or "1024x1536" or "1536x1024" or "auto" The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: "1024x1024" "1024x1536" "1536x1024" "auto" svg: optional { model, reasoning } SVG author model settings. Requires output_format to be svg.
When SVG model selection is enabled, omitting this object uses the
outer response model. Otherwise, the default SVG author is used and
explicit model overrides are unavailable.
model: optional string minLength1 reasoning: optional object or null Reasoning effort for SVG authoring. When omitted, uses the author model’s default in both Images and Responses.
effort: optional "none" or "minimal" or "low" or 4 more Controls how much effort the model should spend on reasoning for this request.
One of the following: "none" "minimal" "low" "medium" "high" "xhigh" "max" LocalShell { type } A tool that allows the model to execute shell commands in a local environment.
type: "local_shell" (default: "local_shell") The type of the local shell tool. Always local_shell.
Shell { type, allowed_callers, environment } A tool that allows the model to execute shell commands.
type: "shell" (default: "shell") The type of the shell tool. Always shell.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" environment: optional ContainerAuto or LocalEnvironment or ContainerReference or null One of the following: ContainerAuto { type, file_ids, memory_limit, 2 more } type: "container_auto" (default: "container_auto") Automatically creates a container for this request
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 skills: optional array of SkillReference or InlineSkill An optional list of skills referenced by id or inline data.
maxItems200 One of the following: SkillReference { skill_id, type, version } skill_id: string The ID of the referenced skill.
minLength1 maxLength64 type: "skill_reference" (default: "skill_reference") References a skill created with the /v1/skills endpoint.
version: optional string Optional skill version. Use a positive integer or ‘latest’. Omit for default.
InlineSkill { description, name, source, type } description: string The description of the skill.
name: string The name of the skill.
source: InlineSkillSource { data, media_type, type } Inline skill payload
type: "inline" (default: "inline") Defines an inline skill for this request.
LocalEnvironment { type, skills } type: "local" (default: "local") Use a local computer environment.
skills: optional array of LocalSkill { description, name, path } An optional list of skills.
maxItems200 description: string The description of the skill.
name: string The name of the skill.
path: string The path to the directory containing the skill.
ContainerReference { container_id, type } container_id: string The ID of the referenced container.
type: "container_reference" (default: "container_reference") References a container created with the /v1/containers endpoint
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
Namespace { description, name, tools, type } Groups function/custom tools under a shared namespace.
description: string A description of the namespace shown to the model.
name: string The namespace name used in tool calls (for example, crm).
minLength1 tools: array of Function or Custom The function/custom tools available inside this namespace.
minItems1 One of the following: Function { name, type, allowed_callers, 6 more } name: string pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 type: "function" (default: "function") allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this function should be deferred and discovered via tool search.
description: optional string or null output_schema: optional map[unknown] or null A JSON Schema describing the JSON value encoded in string outputs for this function tool. This does not describe content-array outputs.
parameters: optional unknown or null strict: optional boolean or null Whether to enforce strict parameter validation. If omitted, Responses attempts to use strict validation when the schema is compatible, and falls back to non-strict validation otherwise.
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
type: "namespace" (default: "namespace") The type of the tool. Always namespace.
ToolSearch { type, description, execution, parameters } Hosted or BYOT tool search configuration for deferred tools.
type: "tool_search" (default: "tool_search") The type of the tool. Always tool_search.
description: optional string or null Description shown to the model for a client-executed tool search tool.
execution: optional "server" or "client" Whether tool search is executed by the server or by the client.
One of the following: "server" "client" parameters: optional unknown or null Parameter schema for a client-executed tool search tool.
WebSearchPreview { type, search_content_types, search_context_size, user_location } This tool searches the web for relevant results to use in a response. Learn more about the web search tool.
type: "web_search_preview" or "web_search_preview_2025_03_11" (default: "web_search_preview") The type of the web search tool. One of web_search_preview or web_search_preview_2025_03_11.
One of the following: "web_search_preview" "web_search_preview_2025_03_11" search_content_types: optional array of "text" or "image" One of the following: "text" "image" search_context_size: optional "low" or "medium" or "high" High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the United States. To avoid this fallback, pass {"type": "approximate"} without location fields. To localize results, provide the relevant location fields.
type: "approximate" (default: "approximate") The type of location approximation. Always approximate.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
ApplyPatch { type, allowed_callers } Allows the assistant to create, delete, or update files using unified diffs.
type: "apply_patch" (default: "apply_patch") The type of the tool. Always apply_patch.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" type: "additional_tools" (default: "additional_tools") The item type. Always additional_tools.
id: optional string or null The unique ID of this additional tools item.
ConfigurationUpdate { type, id, reasoning, text } An update to the conversation’s response configuration. Supply reasoning,
text, or both. Each setting stays in effect until a later update replaces it.
type: "configuration_update" (default: "configuration_update") The item type. Always configuration_update.
id: optional string or null The unique ID of the configuration update item.
reasoning: optional { effort } Updates to reasoning configuration. Only effort is supported.
effort: optional ReasoningEffort or null (default: "medium") The reasoning effort to use for subsequent responses until another
configuration update replaces it.
text: optional { format } The text output configuration for subsequent responses.
format: ResponseFormatText or ResponseFormatTextJSONSchemaConfig Use a JSON schema for structured output or text to return to plain text.
One of the following: ResponseFormatText { type } Default response format. Used to generate text responses.
type: "text" The type of response format being defined. Always text.
ResponseFormatTextJSONSchemaConfig { name, schema, type, 2 more } JSON Schema response format. Used to generate structured JSON responses.
Learn more about Structured Outputs.
name: string The name of the response format. Must be a-z, A-Z, 0-9, or contain
underscores and dashes, with a maximum length of 64.
schema: map[unknown] The schema for the response format, described as a JSON Schema object.
Learn how to build JSON schemas here.
type: "json_schema" The type of response format being defined. Always json_schema.
description: optional string A description of what the response format is for, used by the model to
determine how to respond in the format.
strict: optional boolean or null (default: false) Whether to enable strict schema adherence when generating the output.
If set to true, the model will always follow the exact schema defined
in the schema field. Only a subset of JSON Schema is supported when
strict is true. To learn more, read the Structured Outputs
guide.
Reasoning { id, summary, type, 3 more } A description of the chain of thought used by a reasoning model while generating
a response. Be sure to include these items in your input to the Responses API
for subsequent turns of a conversation if you are manually
managing context.
id: string The unique identifier of the reasoning content.
summary: array of SummaryTextContent { text, type } Reasoning summary content.
text: string A summary of the reasoning output from the model so far.
type: "summary_text" (default: "summary_text") The type of the object. Always summary_text.
type: "reasoning" The type of the object. Always reasoning.
content: optional array of { text, type } Reasoning text content.
text: string The reasoning text from the model.
type: "reasoning_text" (default: "reasoning_text") The type of the reasoning text. Always reasoning_text.
encrypted_content: optional string or null The encrypted content of the reasoning item. This is populated by default
for reasoning items returned by POST /v1/responses and WebSocket
response.create requests.
When streaming, use the completed reasoning item and its
encrypted_content from the response.output_item.done event in
subsequent requests. The encrypted_content in
response.output_item.added may be incomplete. This is especially
important when store is false or when using Zero Data Retention.
status: optional "in_progress" or "completed" or "incomplete" The status of the item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" Compaction { encrypted_content, type, id } A compaction item generated by the v1/responses/compact API.
encrypted_content: string The encrypted content of the compaction summary.
maxLength104857600 type: "compaction" (default: "compaction") The type of the item. Always compaction.
id: optional string or null The ID of the compaction item.
ImageGenerationCall { id, result, status, 9 more } An image generation request made by the model.
id: string The unique ID of the image generation call.
result: string or null The generated image encoded in base64.
status: "in_progress" or "completed" or "generating" or "failed" The status of the image generation call.
One of the following: "in_progress" "completed" "generating" "failed" type: "image_generation_call" The type of the image generation call. Always image_generation_call.
action: optional "generate" or "edit" or "auto" or null The action used for image generation.
One of the following: "generate" "edit" "auto" background: optional "transparent" or "opaque" or "auto" or null The background setting used for generation.
One of the following: "transparent" "opaque" "auto" error: optional { code, message, param, 2 more } Details of a moderation failure, when available.
code: string or null The error code.
message: string A description of the error.
param: string or null The related request parameter.
type: string The error type.
moderation_details: optional { categories, moderation_stage } Public details for a moderation_blocked error, when available.
categories: array of string Public moderation categories. New category values may be added over time.
moderation_stage: "input" or "output" or "unknown" The stage where moderation blocked the image generation call.
One of the following: "input" "output" "unknown" output_format: optional "png" or "webp" or "jpeg" or 2 more or null The output format used for generation.
One of the following: "png" "webp" "jpeg" "gif" "svg" quality: optional "low" or "medium" or "high" or 3 more or null The quality of the image generated by the image generation tool call. One of low, medium, high, xhigh, max, or auto.
One of the following: "low" "medium" "high" "xhigh" "max" "auto" revised_prompt: optional string or null The prompt that was used after any model prompt rewriting.
size: optional string or "1024x1024" or "1024x1536" or "1536x1024" or null The image dimensions as a WIDTHxHEIGHT string, for example 1536x864.
One of the following: string "1024x1024" or "1024x1536" or "1536x1024" The image dimensions as a WIDTHxHEIGHT string, for example 1536x864.
One of the following: "1024x1024" "1024x1536" "1536x1024" svg_usage: optional { input_tokens, model, output_tokens, 2 more } input_tokens: integer model: string output_tokens: integer reasoning_tokens: integer total_tokens: integer CodeInterpreterCall { id, code, container_id, 3 more } A tool call to run code.
id: string The unique ID of the code interpreter tool call.
code: string or null The code to run, or null if not available.
container_id: string The ID of the container used to run the code.
outputs: array of Logs or Image or null The outputs generated by the code interpreter, such as logs or images.
Can be null if no outputs are available.
One of the following: Logs { logs, type } The logs output from the code interpreter.
logs: string The logs output from the code interpreter.
type: "logs" (default: "logs") The type of the output. Always logs.
Image { type, url } The image output from the code interpreter.
type: "image" (default: "image") The type of the output. Always image.
url: string The URL of the image output from the code interpreter.
formaturi status: "in_progress" or "completed" or "incomplete" or 2 more The status of the code interpreter tool call. Valid values are in_progress, completed, incomplete, interpreting, and failed.
One of the following: "in_progress" "completed" "incomplete" "interpreting" "failed" type: "code_interpreter_call" (default: "code_interpreter_call") The type of the code interpreter tool call. Always code_interpreter_call.
LocalShellCall { id, action, call_id, 2 more } A tool call to run a command on the local shell.
id: string The unique ID of the local shell call.
action: { command, env, type, 3 more } Execute a shell command on the server.
command: array of string The command to run.
env: map[string] Environment variables to set for the command.
type: "exec" (default: "exec") The type of the local shell action. Always exec.
timeout_ms: optional integer or null Optional timeout in milliseconds for the command.
user: optional string or null Optional user to run the command as.
working_directory: optional string or null Optional working directory to run the command in.
call_id: string The unique ID of the local shell tool call generated by the model.
status: "in_progress" or "completed" or "incomplete" The status of the local shell call.
One of the following: "in_progress" "completed" "incomplete" type: "local_shell_call" The type of the local shell call. Always local_shell_call.
LocalShellCallOutput { id, output, type, status } The output of a local shell tool call.
id: string The unique ID of the local shell tool call generated by the model.
output: string A JSON string of the output of the local shell tool call.
type: "local_shell_call_output" The type of the local shell tool call output. Always local_shell_call_output.
status: optional "in_progress" or "completed" or "incomplete" or null The status of the item. One of in_progress, completed, or incomplete.
One of the following: "in_progress" "completed" "incomplete" ShellCall { action, call_id, type, 4 more } A tool representing a request to execute one or more shell commands.
action: { commands, max_output_length, timeout_ms } The shell commands and limits that describe how to run the tool call.
commands: array of string Ordered shell commands for the execution environment to run.
max_output_length: optional integer or null Maximum number of UTF-8 characters to capture from combined stdout and stderr output.
timeout_ms: optional integer or null Maximum wall-clock time in milliseconds to allow the shell commands to run.
call_id: string The unique ID of the shell tool call generated by the model.
minLength1 maxLength64 type: "shell_call" (default: "shell_call") The type of the item. Always shell_call.
id: optional string or null The unique ID of the shell tool call. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
environment: optional LocalEnvironment or ContainerReference or null The environment to execute the shell commands in.
One of the following: LocalEnvironment { type, skills } type: "local" (default: "local") Use a local computer environment.
skills: optional array of LocalSkill { description, name, path } An optional list of skills.
maxItems200 description: string The description of the skill.
name: string The name of the skill.
path: string The path to the directory containing the skill.
ContainerReference { container_id, type } container_id: string The ID of the referenced container.
type: "container_reference" (default: "container_reference") References a container created with the /v1/containers endpoint
status: optional "in_progress" or "completed" or "incomplete" or null The status of the shell call. One of in_progress, completed, or incomplete.
One of the following: "in_progress" "completed" "incomplete" ShellCallOutput { call_id, output, type, 4 more } The streamed output items emitted by a shell tool call.
call_id: string The unique ID of the shell tool call generated by the model.
minLength1 maxLength64 output: array of ResponseFunctionShellCallOutputContent { outcome, stderr, stdout } Captured chunks of stdout and stderr output, along with their associated outcomes.
outcome: Timeout or Exit The exit or timeout outcome associated with this shell call.
One of the following: Timeout { type } Indicates that the shell call exceeded its configured time limit.
type: "timeout" (default: "timeout") The outcome type. Always timeout.
Exit { exit_code, type } Indicates that the shell commands finished and returned an exit code.
exit_code: integer The exit code returned by the shell process.
type: "exit" (default: "exit") The outcome type. Always exit.
stderr: string Captured stderr output for the shell call.
maxLength10485760 stdout: string Captured stdout output for the shell call.
maxLength10485760 type: "shell_call_output" (default: "shell_call_output") The type of the item. Always shell_call_output.
id: optional string or null The unique ID of the shell tool call output. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
max_output_length: optional integer or null The maximum number of UTF-8 characters captured for this shell call’s combined output.
status: optional "in_progress" or "completed" or "incomplete" or null The status of the shell call output.
One of the following: "in_progress" "completed" "incomplete" ApplyPatchCall { call_id, operation, status, 3 more } A tool call representing a request to create, delete, or update files using diff patches.
call_id: string The unique ID of the apply patch tool call generated by the model.
minLength1 maxLength64 operation: CreateFile or DeleteFile or UpdateFile The specific create, delete, or update instruction for the apply_patch tool call.
One of the following: CreateFile { diff, path, type } Instruction for creating a new file via the apply_patch tool.
diff: string Unified diff content to apply when creating the file.
maxLength10485760 path: string Path of the file to create relative to the workspace root.
minLength1 type: "create_file" (default: "create_file") The operation type. Always create_file.
DeleteFile { path, type } Instruction for deleting an existing file via the apply_patch tool.
path: string Path of the file to delete relative to the workspace root.
minLength1 type: "delete_file" (default: "delete_file") The operation type. Always delete_file.
UpdateFile { diff, path, type } Instruction for updating an existing file via the apply_patch tool.
diff: string Unified diff content to apply to the existing file.
maxLength10485760 path: string Path of the file to update relative to the workspace root.
minLength1 type: "update_file" (default: "update_file") The operation type. Always update_file.
status: "in_progress" or "completed" The status of the apply patch tool call. One of in_progress or completed.
One of the following: "in_progress" "completed" type: "apply_patch_call" (default: "apply_patch_call") The type of the item. Always apply_patch_call.
id: optional string or null The unique ID of the apply patch tool call. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
ApplyPatchCallOutput { call_id, status, type, 3 more } The streamed output emitted by an apply patch tool call.
call_id: string The unique ID of the apply patch tool call generated by the model.
minLength1 maxLength64 status: "completed" or "failed" The status of the apply patch tool call output. One of completed or failed.
One of the following: "completed" "failed" type: "apply_patch_call_output" (default: "apply_patch_call_output") The type of the item. Always apply_patch_call_output.
id: optional string or null The unique ID of the apply patch tool call output. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
output: optional string or null Optional human-readable log text from the apply patch tool (e.g., patch results or errors).
maxLength10485760 McpListTools { id, server_label, tools, 2 more } A list of tools available on an MCP server.
id: string The unique ID of the list.
server_label: string The label of the MCP server.
tools: array of { input_schema, name, annotations, description } The tools available on the server.
input_schema: unknown The JSON schema describing the tool’s input.
name: string The name of the tool.
annotations: optional unknown or null Additional annotations about the tool.
description: optional string or null The description of the tool.
type: "mcp_list_tools" The type of the item. Always mcp_list_tools.
error: optional string or null Error message if the server could not list tools.
McpApprovalRequest { id, arguments, name, 2 more } A request for human approval of a tool invocation.
id: string The unique ID of the approval request.
arguments: string A JSON string of arguments for the tool.
name: string The name of the tool to run.
server_label: string The label of the MCP server making the request.
type: "mcp_approval_request" The type of the item. Always mcp_approval_request.
McpApprovalResponse { approval_request_id, approve, type, 2 more } A response to an MCP approval request.
approval_request_id: string The ID of the approval request being answered.
approve: boolean Whether the request was approved.
type: "mcp_approval_response" The type of the item. Always mcp_approval_response.
id: optional string or null The unique ID of the approval response
reason: optional string or null Optional reason for the decision.
McpCall { id, arguments, name, 6 more } An invocation of a tool on an MCP server.
id: string The unique ID of the tool call.
arguments: string A JSON string of the arguments passed to the tool.
name: string The name of the tool that was run.
server_label: string The label of the MCP server running the tool.
type: "mcp_call" The type of the item. Always mcp_call.
approval_request_id: optional string or null Unique identifier for the MCP tool call approval request.
Include this value in a subsequent mcp_approval_response input to approve or reject the corresponding tool call.
error: optional McpToolCallError or null The error from the tool call, if any.
output: optional string or null The output from the tool call.
status: optional "in_progress" or "completed" or "incomplete" or 2 more The status of the tool call. One of in_progress, completed, incomplete, calling, or failed.
One of the following: "in_progress" "completed" "incomplete" "calling" "failed" CustomToolCallOutput { call_id, output, type, 2 more } The output of a custom tool call from your code, being sent back to the model.
call_id: string The call ID, used to map this custom tool call output to a custom tool call.
output: string or array of ResponseInputText or ResponseInputImage or ResponseInputFile The output from the custom tool call generated by your code.
Can be a string or an list of output content.
One of the following: StringOutput = string A string of the output of the custom tool call.
OutputContentList = array of ResponseInputText or ResponseInputImage or ResponseInputFile Text, image, or file output of the custom tool call.
One of the following: ResponseInputText { text, type, prompt_cache_breakpoint } A text input to the model.
text: string The text input to the model.
type: "input_text" (default: "input_text") The type of the input item. Always input_text.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputImage { detail, type, file_id, 2 more } An image input to the model. Learn about image inputs.
detail: ImageDetail The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
type: "input_image" (default: "input_image") The type of the input item. Always input_image.
file_id: optional string or null The ID of the file to be sent to the model.
image_url: optional string or null The URL of the image to be sent to the model. A fully qualified URL or base64 encoded image in a data URL.
formaturi prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputFile { type, detail, file_data, 4 more } A file input to the model.
type: "input_file" (default: "input_file") The type of the input item. Always input_file.
detail: optional "auto" or "low" or "high" The detail level of the file to be sent to the model. Use auto to let the system select the detail level; for GPT-5.6 and later models, auto uses high-quality rendering, which may increase input token usage. Use low for lower-cost rendering, or high to render the file at higher quality. Defaults to auto.
One of the following: "auto" "low" "high" file_data: optional string The content of the file to be sent to the model.
file_id: optional string or null The ID of the file to be sent to the model.
file_url: optional string The URL of the file to be sent to the model.
formaturi filename: optional string The name of the file to be sent to the model.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" type: "custom_tool_call_output" The type of the custom tool call output. Always custom_tool_call_output.
id: optional string The unique ID of the custom tool call output in the OpenAI platform.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
CustomToolCall { call_id, input, name, 5 more } A call to a custom tool created by the model.
call_id: string An identifier used to map this custom tool call to a tool call output.
input: string The input for the custom tool call generated by the model.
name: string The name of the custom tool being called.
type: "custom_tool_call" The type of the custom tool call. Always custom_tool_call.
id: optional string The unique ID of the custom tool call in the OpenAI platform.
async: optional boolean Whether the custom tool call runs asynchronously.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
type: "program" (default: "program") namespace: optional string The namespace of the custom tool being called.
CompactionTrigger { type, id } Compacts the current context. Must be the final input item.
type: "compaction_trigger" (default: "compaction_trigger") The type of the item. Always compaction_trigger.
id: optional string or null The unique ID of this compaction trigger.
ItemReference { id, type } An internal identifier for an item to reference.
id: string The ID of the item to reference.
type: optional "item_reference" or null (default: "item_reference") The type of item to reference. Always item_reference.
Program { id, call_id, code, 2 more } id: string The unique ID of this program item.
call_id: string The stable call ID of the program item.
minLength1 maxLength64 code: string The JavaScript source executed by programmatic tool calling.
maxLength10485760 fingerprint: string Opaque program replay fingerprint that must be round-tripped.
maxLength10485760 type: "program" (default: "program") The item type. Always program.
ProgramOutput { id, call_id, result, 2 more } id: string The unique ID of this program output item.
call_id: string The call ID of the program item.
minLength1 maxLength64 result: string The result produced by the program item.
maxLength10485760 status: "completed" or "incomplete" The terminal status of the program output.
One of the following: "completed" "incomplete" type: "program_output" (default: "program_output") The item type. Always program_output.
type: "response.item.create" (default: "response.item.create") The Live client event type. Always response.item.create.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512
ResponsesDelegationConfig object { model, instructions, max_output_tokens, 6 more } Model, prompt, and tool settings for tasks delegated by the Live session to a Responses backend.
ResponsesDelegationConfig object { model, instructions, max_output_tokens, 6 more }
Model, prompt, and tool settings for tasks delegated by the Live session to a Responses backend.
model: string The model used for server-owned Responses delegations.
instructions: optional string or null Instructions for the delegated Responses model, separate from Live instructions. See backend prompting.
max_output_tokens: optional integer or null Maximum number of output tokens for each delegated response.
minimum16 parallel_tool_calls: optional boolean or null Whether the delegated Responses model may request multiple tool calls in a single response.
reasoning: optional object or null Reasoning settings passed to each delegated Responses request.
effort: optional "none" or "minimal" or "low" or 3 more or null How much reasoning effort the delegated Responses model should use. Supported values depend on the backend model.
One of the following: "none" "minimal" "low" "medium" "high" "xhigh" summary: optional "concise" or "detailed" or "auto" or null The reasoning summary to request from the delegated Responses model, when supported.
One of the following: "concise" "detailed" "auto" service_tier: optional "auto" or "default" or "fast_tier_temp_pilot" or 3 more or null Service tier for delegated Responses requests.
One of the following: "auto" "default" "fast_tier_temp_pilot" "flex" "priority" "ultrafast" text: optional object or null Text generation settings passed to each delegated Responses request.
verbosity: optional "low" or "medium" or "high" or null The amount of detail in text generated by the Responses backend. This does not configure the Live model’s spoken delivery.
One of the following: "low" "medium" "high" tool_choice: optional "auto" or "none" or "required" or Function or Mcp or 11 more Controls which tool the Responses backend uses when handling a task delegated by the Live model.
One of the following: LiveToolChoiceEnum = "auto" or "none" or "required" One of the following: "auto" "none" "required" Function { name, type } name: string The name of the function tool to call.
type: "function" (default: "function") The tool to call. Always function.
Mcp { server_label, type, name } server_label: string The label of the MCP server to call.
type: "mcp" (default: "mcp") The tool to call. Always mcp.
name: optional string or null The name of the MCP tool to call. If omitted, the server may choose a default.
FileSearch { type } type: "file_search" (default: "file_search") The tool to call. Always file_search.
WebSearch { type } type: "web_search" (default: "web_search") The tool to call. Always web_search.
WebSearchPreview { type } type: "web_search_preview" (default: "web_search_preview") The tool to call. Always web_search_preview.
ImageGeneration { type } type: "image_generation" (default: "image_generation") The tool to call. Always image_generation.
Computer { type } type: "computer" (default: "computer") The tool to call. Always computer.
CodeInterpreter { type } type: "code_interpreter" (default: "code_interpreter") The tool to call. Always code_interpreter.
ProgrammaticToolCalling { type } type: "programmatic_tool_calling" (default: "programmatic_tool_calling") The tool to call. Always programmatic_tool_calling.
Shell { type } Forces the model to call the shell tool when a tool call is required.
type: "shell" (default: "shell") The tool to call. Always shell.
Custom { name, type } name: string The name of the custom tool to call.
type: "custom" (default: "custom") The tool to call. Always custom.
ApplyPatch { type } Forces the model to call the apply_patch tool when executing a tool call.
type: "apply_patch" (default: "apply_patch") The tool to call. Always apply_patch.
LiveAllowedToolsChoiceParam { tools, type, mode } tools: array of Function or Mcp or FileSearch or 9 more The tools that the delegated Responses model may call.
minItems1 maxItems128 One of the following: Function { name, type } name: string The name of the function tool to call.
type: "function" (default: "function") The tool to call. Always function.
Mcp { server_label, type, name } server_label: string The label of the MCP server to call.
type: "mcp" (default: "mcp") The tool to call. Always mcp.
name: optional string or null The name of the MCP tool to call. If omitted, the server may choose a default.
FileSearch { type } type: "file_search" (default: "file_search") The tool to call. Always file_search.
WebSearch { type } type: "web_search" (default: "web_search") The tool to call. Always web_search.
WebSearchPreview { type } type: "web_search_preview" (default: "web_search_preview") The tool to call. Always web_search_preview.
ImageGeneration { type } type: "image_generation" (default: "image_generation") The tool to call. Always image_generation.
Computer { type } type: "computer" (default: "computer") The tool to call. Always computer.
CodeInterpreter { type } type: "code_interpreter" (default: "code_interpreter") The tool to call. Always code_interpreter.
ProgrammaticToolCalling { type } type: "programmatic_tool_calling" (default: "programmatic_tool_calling") The tool to call. Always programmatic_tool_calling.
Shell { type } Forces the model to call the shell tool when a tool call is required.
type: "shell" (default: "shell") The tool to call. Always shell.
Custom { name, type } name: string The name of the custom tool to call.
type: "custom" (default: "custom") The tool to call. Always custom.
ApplyPatch { type } Forces the model to call the apply_patch tool when executing a tool call.
type: "apply_patch" (default: "apply_patch") The tool to call. Always apply_patch.
type: "allowed_tools" (default: "allowed_tools") The tool choice type. Always allowed_tools.
mode: optional "none" or "auto" or "required" or null How to select a tool from the allowed set.
One of the following: "none" "auto" "required" tools: optional array of FunctionTool or WebSearch or FileSearch or 10 more Tools available to the Responses backend while it handles tasks delegated by the Live model.
One of the following: FunctionTool { name, type, description, 2 more } A function tool available to the Responses backend when the Live model delegates a task.
name: string The name the delegated Responses model uses when calling this function.
type: "function" (default: "function") The tool type. Always function.
description: optional string or null What the function does and when the delegated Responses model should call it.
parameters: optional map[unknown] or null A JSON Schema object describing the arguments accepted by the function.
strict: optional boolean or null Whether the delegated Responses model must follow the function’s parameter schema exactly.
WebSearch { type } A web search tool available to the Live session’s Responses backend.
type: "web_search" (default: "web_search") The tool type. Always web_search.
FileSearch { type } type: "file_search" (default: "file_search") CodeInterpreter { type } type: "code_interpreter" (default: "code_interpreter") Shell { type, environment } A Responses shell tool. Use a hosted container or return local shell results with response.item.create. Domain secrets are not supported.
type: "shell" (default: "shell") environment: optional ContainerAuto or ContainerReference or Local or null One of the following: ContainerAuto { type, file_ids, memory_limit, 2 more } type: "container_auto" (default: "container_auto") Automatically creates a container for this request
file_ids: optional array of string or null An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional Disabled or Allowlist or null Network access policy for the container.
One of the following: Disabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
Allowlist { allowed_domains, type } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
skills: optional array of SkillReference or Inline or null An optional list of skills referenced by id or inline data.
maxItems200 One of the following: SkillReference { skill_id, type, version } skill_id: string The ID of the referenced skill.
minLength1 maxLength64 type: "skill_reference" (default: "skill_reference") References a skill created with the /v1/skills endpoint.
version: optional string or null Optional skill version. Use a positive integer or ‘latest’. Omit for default.
Inline { description, name, source, type } description: string The description of the skill.
name: string The name of the skill.
source: { data, media_type, type } Inline skill payload
data: string Base64-encoded skill zip bundle.
minLength1 maxLength70254592 media_type: "application/zip" (default: "application/zip") The media type of the inline skill payload. Must be application/zip.
type: "base64" (default: "base64") The type of the inline skill source. Must be base64.
type: "inline" (default: "inline") Defines an inline skill for this request.
ContainerReference { container_id, type } container_id: string The ID of the referenced container.
type: "container_reference" (default: "container_reference") References a container created with the /v1/containers endpoint
Local { type, skills } type: "local" (default: "local") Use a local computer environment.
skills: optional array of object or null An optional list of skills.
maxItems200 description: string The description of the skill.
name: string The name of the skill.
path: string The path to the directory containing the skill.
ImageGeneration { type } type: "image_generation" (default: "image_generation") Mcp { type } type: "mcp" (default: "mcp") Custom { type } type: "custom" (default: "custom") Namespace { type } type: "namespace" (default: "namespace") ToolSearch { type } type: "tool_search" (default: "tool_search") ProgrammaticToolCalling { type } type: "programmatic_tool_calling" (default: "programmatic_tool_calling") Computer { type } type: "computer" (default: "computer") ApplyPatch { type } type: "apply_patch" (default: "apply_patch")
ResponsesDelegationUpdateConfig object { instructions, max_output_tokens, model, 6 more } Updates to the Responses backend of an existing Live session. Omitted settings retain their current values.
ResponsesDelegationUpdateConfig object { instructions, max_output_tokens, model, 6 more }
Updates to the Responses backend of an existing Live session. Omitted settings retain their current values.
instructions: optional string or null Instructions for the delegated Responses model, separate from Live instructions. See backend prompting.
max_output_tokens: optional integer or null Maximum number of output tokens for each delegated response.
minimum16 model: optional string The Responses backend model to use for subsequent delegated requests. Omit to keep the current backend model.
parallel_tool_calls: optional boolean or null Whether the delegated Responses model may request multiple tool calls in a single response.
reasoning: optional object or null Reasoning settings passed to each delegated Responses request.
effort: optional "none" or "minimal" or "low" or 3 more or null How much reasoning effort the delegated Responses model should use. Supported values depend on the backend model.
One of the following: "none" "minimal" "low" "medium" "high" "xhigh" summary: optional "concise" or "detailed" or "auto" or null The reasoning summary to request from the delegated Responses model, when supported.
One of the following: "concise" "detailed" "auto" service_tier: optional "auto" or "default" or "fast_tier_temp_pilot" or 3 more or null Service tier for delegated Responses requests.
One of the following: "auto" "default" "fast_tier_temp_pilot" "flex" "priority" "ultrafast" text: optional object or null Text generation settings passed to each delegated Responses request.
verbosity: optional "low" or "medium" or "high" or null The amount of detail in text generated by the Responses backend. This does not configure the Live model’s spoken delivery.
One of the following: "low" "medium" "high" tool_choice: optional "auto" or "none" or "required" or Function or Mcp or 11 more Controls which tool the Responses backend uses when handling a task delegated by the Live model.
One of the following: LiveToolChoiceEnum = "auto" or "none" or "required" One of the following: "auto" "none" "required" Function { name, type } name: string The name of the function tool to call.
type: "function" (default: "function") The tool to call. Always function.
Mcp { server_label, type, name } server_label: string The label of the MCP server to call.
type: "mcp" (default: "mcp") The tool to call. Always mcp.
name: optional string or null The name of the MCP tool to call. If omitted, the server may choose a default.
FileSearch { type } type: "file_search" (default: "file_search") The tool to call. Always file_search.
WebSearch { type } type: "web_search" (default: "web_search") The tool to call. Always web_search.
WebSearchPreview { type } type: "web_search_preview" (default: "web_search_preview") The tool to call. Always web_search_preview.
ImageGeneration { type } type: "image_generation" (default: "image_generation") The tool to call. Always image_generation.
Computer { type } type: "computer" (default: "computer") The tool to call. Always computer.
CodeInterpreter { type } type: "code_interpreter" (default: "code_interpreter") The tool to call. Always code_interpreter.
ProgrammaticToolCalling { type } type: "programmatic_tool_calling" (default: "programmatic_tool_calling") The tool to call. Always programmatic_tool_calling.
Shell { type } Forces the model to call the shell tool when a tool call is required.
type: "shell" (default: "shell") The tool to call. Always shell.
Custom { name, type } name: string The name of the custom tool to call.
type: "custom" (default: "custom") The tool to call. Always custom.
ApplyPatch { type } Forces the model to call the apply_patch tool when executing a tool call.
type: "apply_patch" (default: "apply_patch") The tool to call. Always apply_patch.
LiveAllowedToolsChoiceParam { tools, type, mode } tools: array of Function or Mcp or FileSearch or 9 more The tools that the delegated Responses model may call.
minItems1 maxItems128 One of the following: Function { name, type } name: string The name of the function tool to call.
type: "function" (default: "function") The tool to call. Always function.
Mcp { server_label, type, name } server_label: string The label of the MCP server to call.
type: "mcp" (default: "mcp") The tool to call. Always mcp.
name: optional string or null The name of the MCP tool to call. If omitted, the server may choose a default.
FileSearch { type } type: "file_search" (default: "file_search") The tool to call. Always file_search.
WebSearch { type } type: "web_search" (default: "web_search") The tool to call. Always web_search.
WebSearchPreview { type } type: "web_search_preview" (default: "web_search_preview") The tool to call. Always web_search_preview.
ImageGeneration { type } type: "image_generation" (default: "image_generation") The tool to call. Always image_generation.
Computer { type } type: "computer" (default: "computer") The tool to call. Always computer.
CodeInterpreter { type } type: "code_interpreter" (default: "code_interpreter") The tool to call. Always code_interpreter.
ProgrammaticToolCalling { type } type: "programmatic_tool_calling" (default: "programmatic_tool_calling") The tool to call. Always programmatic_tool_calling.
Shell { type } Forces the model to call the shell tool when a tool call is required.
type: "shell" (default: "shell") The tool to call. Always shell.
Custom { name, type } name: string The name of the custom tool to call.
type: "custom" (default: "custom") The tool to call. Always custom.
ApplyPatch { type } Forces the model to call the apply_patch tool when executing a tool call.
type: "apply_patch" (default: "apply_patch") The tool to call. Always apply_patch.
type: "allowed_tools" (default: "allowed_tools") The tool choice type. Always allowed_tools.
mode: optional "none" or "auto" or "required" or null How to select a tool from the allowed set.
One of the following: "none" "auto" "required" tools: optional array of FunctionTool or WebSearch or FileSearch or 10 more Tools available to the Responses backend while it handles tasks delegated by the Live model.
One of the following: FunctionTool { name, type, description, 2 more } A function tool available to the Responses backend when the Live model delegates a task.
name: string The name the delegated Responses model uses when calling this function.
type: "function" (default: "function") The tool type. Always function.
description: optional string or null What the function does and when the delegated Responses model should call it.
parameters: optional map[unknown] or null A JSON Schema object describing the arguments accepted by the function.
strict: optional boolean or null Whether the delegated Responses model must follow the function’s parameter schema exactly.
WebSearch { type } A web search tool available to the Live session’s Responses backend.
type: "web_search" (default: "web_search") The tool type. Always web_search.
FileSearch { type } type: "file_search" (default: "file_search") CodeInterpreter { type } type: "code_interpreter" (default: "code_interpreter") Shell { type, environment } A Responses shell tool. Use a hosted container or return local shell results with response.item.create. Domain secrets are not supported.
type: "shell" (default: "shell") environment: optional ContainerAuto or ContainerReference or Local or null One of the following: ContainerAuto { type, file_ids, memory_limit, 2 more } type: "container_auto" (default: "container_auto") Automatically creates a container for this request
file_ids: optional array of string or null An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional Disabled or Allowlist or null Network access policy for the container.
One of the following: Disabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
Allowlist { allowed_domains, type } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
skills: optional array of SkillReference or Inline or null An optional list of skills referenced by id or inline data.
maxItems200 One of the following: SkillReference { skill_id, type, version } skill_id: string The ID of the referenced skill.
minLength1 maxLength64 type: "skill_reference" (default: "skill_reference") References a skill created with the /v1/skills endpoint.
version: optional string or null Optional skill version. Use a positive integer or ‘latest’. Omit for default.
Inline { description, name, source, type } description: string The description of the skill.
name: string The name of the skill.
source: { data, media_type, type } Inline skill payload
data: string Base64-encoded skill zip bundle.
minLength1 maxLength70254592 media_type: "application/zip" (default: "application/zip") The media type of the inline skill payload. Must be application/zip.
type: "base64" (default: "base64") The type of the inline skill source. Must be base64.
type: "inline" (default: "inline") Defines an inline skill for this request.
ContainerReference { container_id, type } container_id: string The ID of the referenced container.
type: "container_reference" (default: "container_reference") References a container created with the /v1/containers endpoint
Local { type, skills } type: "local" (default: "local") Use a local computer environment.
skills: optional array of object or null An optional list of skills.
maxItems200 description: string The description of the skill.
name: string The name of the skill.
path: string The path to the directory containing the skill.
ImageGeneration { type } type: "image_generation" (default: "image_generation") Mcp { type } type: "mcp" (default: "mcp") Custom { type } type: "custom" (default: "custom") Namespace { type } type: "namespace" (default: "namespace") ToolSearch { type } type: "tool_search" (default: "tool_search") ProgrammaticToolCalling { type } type: "programmatic_tool_calling" (default: "programmatic_tool_calling") Computer { type } type: "computer" (default: "computer") ApplyPatch { type } type: "apply_patch" (default: "apply_patch")
ServerEvent = SessionStartedEvent { event_id, session, type, client_event_id } or SessionUpdatedEvent { event_id, session, type, client_event_id } or InputAudioMutedEvent { event_id, type, client_event_id } or 24 moreServer events for Live. Response lifecycle events are wrapped inside response.event; dispatch the nested event by its full type and tolerate new response event types. Follow the Live prompting guide when designing the conversation and delegation policy.
ServerEvent = SessionStartedEvent { event_id, session, type, client_event_id } or SessionUpdatedEvent { event_id, session, type, client_event_id } or InputAudioMutedEvent { event_id, type, client_event_id } or 24 more
Server events for Live. Response lifecycle events are wrapped inside response.event; dispatch the nested event by its full type and tolerate new response event types. Follow the Live prompting guide when designing the conversation and delegation policy.
One of the following: SessionStartedEvent { event_id, session, type, client_event_id } Returned when a Live session has started. Contains the resolved session configuration, including server defaults.
event_id: string The unique ID of the Live server event.
session: SessionResource { id, expires_at, model, 9 more } The resolved Live session configuration and server-assigned session metadata.
type: "session.started" (default: "session.started") The event type, always session.started.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
SessionUpdatedEvent { event_id, session, type, client_event_id } Returned when a Live session update is accepted. Contains the resolved session configuration after the update.
event_id: string The unique ID of the Live server event.
session: SessionResource { id, expires_at, model, 9 more } The resolved Live session configuration and server-assigned session metadata.
type: "session.updated" (default: "session.updated") The event type, always session.updated.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InputAudioMutedEvent { event_id, type, client_event_id } Acknowledges a session.input_audio.mute command. An independent input mute sends silence to the model and preserves sideband audio reflection. Fixture playback does not emit this event.
event_id: string The unique ID of the Live server event.
type: "session.input_audio.muted" (default: "session.input_audio.muted") The event type, always session.input_audio.muted.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InputAudioUnmutedEvent { event_id, type, client_event_id } Acknowledges a session.input_audio.unmute command, removing the independent input mute. An active audio fixture still pauses inference until playback finishes. Fixture playback does not emit this event.
event_id: string The unique ID of the Live server event.
type: "session.input_audio.unmuted" (default: "session.input_audio.unmuted") The event type, always session.input_audio.unmuted.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
OutputAudioFixturePlayingEvent { event_id, type, client_event_id } The existing playback path has accepted the validated audio fixture. This acknowledgment does not identify when the listener receives the first sample. The optional client_event_id echoes the play command.
event_id: string The unique ID of the Live server event.
type: "session.output_audio.fixture.playing" (default: "session.output_audio.fixture.playing") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
OutputAudioFixturePlayedEvent { event_id, type, client_event_id } The complete audio fixture has drained from the server mixer and normal inference has resumed. Transport buffers may still contain audio; this event does not prove listener-side completion. The optional client_event_id echoes the play command. This event is not emitted for canceled or failed playback.
event_id: string The unique ID of the Live server event.
type: "session.output_audio.fixture.played" (default: "session.output_audio.fixture.played") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InstructionsAppendedEvent { end_ms, event_id, start_ms, 2 more } Returned when a session.instructions.append command is accepted into the Live session timeline. Acknowledges the appended instructions without guaranteeing that the model has acted on them.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.instructions.appended" (default: "session.instructions.appended") The event type, always session.instructions.appended.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
ThinkingAppendedEvent { end_ms, event_id, start_ms, 2 more } Returned when a session.thinking.append command is accepted into the Live session timeline. Acknowledges the added reasoning context without guaranteeing any spoken output.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.thinking.appended" (default: "session.thinking.appended") The event type, always session.thinking.appended.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
CommentaryAppendedEvent { end_ms, event_id, start_ms, 2 more } Returned when a session.commentary.append command is accepted into the Live session timeline. Acknowledges the added commentary without guaranteeing exact wording or completed audio playback.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.commentary.appended" (default: "session.commentary.appended") The event type, always session.commentary.appended.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
SessionInputAudioAppend { audio, type } Input audio received from the primary transport and reflected to a Live sideband connection before model-input muting.
audio: string Base64-encoded raw mono PCM16LE at 24 kHz received from the primary transport, reflected to the sideband before model-input muting. This server event uses the same audio key as the client command, but is not an acknowledgment of it.
type: "session.input_audio.append" (default: "session.input_audio.append") The event type, always session.input_audio.append.
OutputAudioDeltaEvent { delta, type, end_ms, start_ms } An audio chunk generated by the Live model. Decode and play primary WebSocket chunks in delivery order using the configured session audio format. Sideband connections receive reflected output audio with timestamps.
delta: string Base64-encoded raw audio. Primary WebSocket events use the session’s configured format; reflected sideband events use mono PCM16LE at 24 kHz.
type: "session.output_audio.delta" (default: "session.output_audio.delta") The event type, always session.output_audio.delta.
end_ms: optional integer Exclusive session-relative end in milliseconds. Required on reflected sideband events; omitted on the primary WebSocket. Dropped output frames leave gaps between reflected ranges.
start_ms: optional integer Inclusive session-relative start in milliseconds. Required on reflected sideband events; omitted on the primary WebSocket.
InputTranscriptDeltaEvent { delta, end_ms, event_id, 3 more } A transcript fragment for user input audio in the Live session. Accumulate fragments in delivery order; these events do not define complete turns or include a transcript-done event.
delta: string The transcript text fragment for the audio in this time range. Append fragments in delivery order to build the transcript.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.input_transcript.delta" (default: "session.input_transcript.delta") The event type, always session.input_transcript.delta.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
OutputTranscriptDeltaEvent { delta, end_ms, event_id, 3 more } A transcript fragment for assistant output audio in the Live session. Accumulate fragments in delivery order; these events do not define complete turns or include a transcript-done event.
delta: string The transcript text fragment for the audio in this time range. Append fragments in delivery order to build the transcript.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.output_transcript.delta" (default: "session.output_transcript.delta") The event type, always session.output_transcript.delta.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
DelegationCreatedEvent { delegation, event_id, offset_ms, 2 more } Returned when the Live model delegates work to your application, Responses, or Decisions. Contains delegation metadata and the position on the session timeline where the work was delegated.
delegation: { id, target, type, response_id } The delegated work identifier and destination. This object contains metadata, not the task text.
id: string The unique ID of the delegation. Use this as delegation_id when replying to client-owned work or correlating backend events.
target: "client" or "responses" or "decisions" Where the Live model delegated the work: client, responses, or decisions.
One of the following: "client" or "responses" or "decisions" Where the Live model delegated the work: client, responses, or decisions.
One of the following: "client" "responses" "decisions" type: "delegation" (default: "delegation") The object type, always delegation.
response_id: optional string The ID of the Responses API response associated with a Responses delegation. Omitted for client delegations.
event_id: string The unique ID of the Live server event.
offset_ms: integer The position on the Live session timeline where the delegation was created, in milliseconds from the beginning of the session.
type: "session.delegation.created" (default: "session.delegation.created") The event type, always session.delegation.created.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
DecisionEvent { delegation_id, event, event_id, 4 more } A Decision API event for a Live delegation or transcript trigger. For a trigger, source_event_id identifies the transcript delta that caused this evaluation.
delegation_id: string or null The Live delegation ID, or null for a transcript trigger.
event: { decision, type } decision: Decision { answers, model, usage } The native Decision API response, including its model, ordered answers, and token usage.
type: "decision.completed" (default: "decision.completed") event_id: string The unique ID of this Live server event.
source_event_id: string or null The event_id of the transcript delta that fired this trigger, or null for a delegation.
source_event_type: "session.input_transcript.delta" or "session.output_transcript.delta" or null One of the following: "session.input_transcript.delta" or "session.output_transcript.delta" One of the following: "session.input_transcript.delta" "session.output_transcript.delta" trigger_id: string or null The Live-assigned ID of this trigger firing, or null for a delegation.
type: "decision.event" (default: "decision.event") DecisionItemCreatedEvent { event_id, item, type, client_event_id } An input item was appended for future decisions.
event_id: string The unique ID of the Live server event.
item: Message or Message or Message or 5 more One of the following: Message { id, content, role, type } id: string content: string or array of InputText or InputImage or InputJSON The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of InputText or InputImage or InputJSON One of the following: InputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") role: "user" (default: "user") The message role. Always user.
type: optional "message" (default: "message") Message { id, content, role, type } id: string content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "developer" (default: "developer") type: optional "message" (default: "message") Message { id, content, role, type } id: string content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "system" (default: "system") type: optional "message" (default: "message") Message { id, content, role, 2 more } id: string content: string or array of OutputText or Refusal The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of OutputText or Refusal One of the following: OutputText { text, type } text: string The text content.
maxLength10485760 type: "output_text" (default: "output_text") Refusal { refusal, type } refusal: string The refusal text.
maxLength10485760 type: "refusal" (default: "refusal") role: "assistant" (default: "assistant") phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer). For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" type: optional "message" (default: "message") FunctionCall { id, arguments, call_id, 4 more } id: string arguments: string The function arguments as a JSON string.
call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 name: string pattern^[a-zA-Z0-9_.-]+$ minLength1 maxLength128 type: "function_call" (default: "function_call") The item type. Always function_call.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" FunctionCallOutput { id, call_id, output, 2 more } id: string call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "function_call_output" (default: "function_call_output") The type of the function tool call output. Always function_call_output.
status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCall { id, call_id, input, 4 more } id: string call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 input: string The input to the custom tool call.
name: string The name of the custom tool to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 type: "custom_tool_call" (default: "custom_tool_call") The item type. Always custom_tool_call.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCallOutput { id, call_id, output, 2 more } id: string call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "custom_tool_call_output" (default: "custom_tool_call_output") status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" type: "decision.item.created" (default: "decision.item.created") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
DecisionItemDeletedEvent { event_id, item_id, type, client_event_id } An input item was removed from future decisions.
event_id: string The unique ID of the Live server event.
item_id: string type: "decision.item.deleted" (default: "decision.item.deleted") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
ResponseEvent { event, event_id, type, 2 more } A streaming Responses API event from a backend delegated to by the Live session. Use the outer delegation_id to associate the nested stream with its Live delegation.
event: map[unknown] The nested Responses streaming event. Dispatch on its type field. Response lifecycle snapshots omit input and clear instructions, tools, and output to keep messages small; consume granular output events for the generated content.
event_id: string The unique ID of the Live server event.
type: "response.event" (default: "response.event") The event type, always response.event.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
delegation_id: optional string or null The Live delegation associated with the nested Responses event. May be null or omitted when the event cannot be correlated with a delegation.
SessionUsageUpdatedEvent { event_id, type, usage, 2 more } Reports cumulative Live audio usage and, when available, the most recent context-window usage. Delegated Responses token usage is reported separately in response.event events.
event_id: string The unique ID of the Live server event.
type: "session.usage.updated" (default: "session.usage.updated") The event type, always session.usage.updated.
usage: SessionUsage { seconds } The cumulative Live audio usage so far.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
context_window: optional { usage_ratio } The latest measured Live context-window usage. Omitted when the context limit is unknown.
usage_ratio: number The latest active context token count divided by the Live model context limit. Can decrease after compaction and may lag between measured audio frames.
SessionClosedEvent { event_id, reason, session, 3 more } Returned after the Live session finishes finalizing, with the close reason, final session snapshot, and cumulative audio usage. A connection closing without this event does not confirm successful finalization.
event_id: string The unique ID of the Live server event.
reason: "close_requested" or "expired" or "content" or 2 more Why the Live session ended: close_requested for an application close or hangup request, expired for the session duration limit, content for a safety filter, remote_hangup for a graceful remote disconnect, or connection_lost for an unexpected primary or upstream disconnection.
One of the following: "close_requested" or "expired" or "content" or 2 more Why the Live session ended: close_requested for an application close or hangup request, expired for the session duration limit, content for a safety filter, remote_hangup for a graceful remote disconnect, or connection_lost for an unexpected primary or upstream disconnection.
One of the following: "close_requested" "expired" "content" "remote_hangup" "connection_lost" session: SessionResource { id, expires_at, model, 9 more } The resolved Live session configuration and server-assigned session metadata.
type: "session.closed" (default: "session.closed") The event type, always session.closed.
usage: SessionUsage { seconds } The final cumulative Live audio usage after session finalization.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
ErrorEvent { error, event_id, type, client_event_id } Reports an error in the Live session, such as an invalid client command. Use error.client_event_id, when present, to identify the command that caused the error.
error: Error { code, message, type, 2 more } Details of the Live error and the client command that caused it, when known.
event_id: string The unique ID of the Live server event.
type: "error" (default: "error") The event type, always error.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InfoEvent { code, event_id, message, 2 more } An informational notice about the Live session, such as the event permissions applied to a frontend data channel.
code: string A machine-readable code for the notice, such as data_channel_permissions.
event_id: string The unique ID of the Live server event.
message: string A human-readable explanation of the Live session notice.
type: "info" (default: "info") The event type, always info.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
TransportDtmfReceived { event, event_id, type } A SIP DTMF keypress received from the caller. Delivered only to sideband observers.
event: string pattern^[0-9A-D*#]$ minLength1 maxLength1 event_id: string type: "transport.dtmf.received" (default: "transport.dtmf.received") TransportDtmfSent { event, event_id, type, client_event_id } A SIP DTMF keypress successfully sent to the SIP trunk. Delivered only to sideband observers; this is not a client command.
event: string pattern^[0-9A-D*#]$ minLength1 maxLength1 event_id: string type: "transport.dtmf.sent" (default: "transport.dtmf.sent") client_event_id: optional string The event_id of the client command, when supplied.
TransportRinging { event_id, session_id, type } The outbound SIP provider leg is ringing or providing early media. Delivered only to sideband observers.
event_id: string session_id: string The canonical Live session ID.
type: "transport.ringing" (default: "transport.ringing") TransportAnswered { event_id, session_id, type } The outbound SIP provider leg answered and media is established. Delivered only to sideband observers.
event_id: string session_id: string The canonical Live session ID.
type: "transport.answered" (default: "transport.answered") TransportFailed { error, event_id, session_id, type } An asynchronous outbound SIP setup failure. Delivered only to sideband observers.
error: { code, message, type, param } code: string The call setup failure code.
message: string type: "call_error" (default: "call_error") param: optional string The parameter related to the error, if any. Empty when no parameter applies.
event_id: string session_id: string The canonical Live session ID.
type: "transport.failed" (default: "transport.failed")
SessionClosedEvent object { event_id, reason, session, 3 more } Returned after the Live session finishes finalizing, with the close reason, final session snapshot, and cumulative audio usage. A connection closing without this event does not confirm successful finalization.
SessionClosedEvent object { event_id, reason, session, 3 more }
Returned after the Live session finishes finalizing, with the close reason, final session snapshot, and cumulative audio usage. A connection closing without this event does not confirm successful finalization.
event_id: string The unique ID of the Live server event.
reason: "close_requested" or "expired" or "content" or 2 more Why the Live session ended: close_requested for an application close or hangup request, expired for the session duration limit, content for a safety filter, remote_hangup for a graceful remote disconnect, or connection_lost for an unexpected primary or upstream disconnection.
One of the following: "close_requested" or "expired" or "content" or 2 more Why the Live session ended: close_requested for an application close or hangup request, expired for the session duration limit, content for a safety filter, remote_hangup for a graceful remote disconnect, or connection_lost for an unexpected primary or upstream disconnection.
One of the following: "close_requested" "expired" "content" "remote_hangup" "connection_lost" session: SessionResource { id, expires_at, model, 9 more } The resolved Live session configuration and server-assigned session metadata.
type: "session.closed" (default: "session.closed") The event type, always session.closed.
usage: SessionUsage { seconds } The final cumulative Live audio usage after session finalization.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
SessionConfig object { model, audio, client, 6 more } Initial configuration for a Live session, including its model, conversation instructions, audio, and delegated task handling.
SessionConfig object { model, audio, client, 6 more }
Initial configuration for a Live session, including its model, conversation instructions, audio, and delegated task handling.
model: string or "gpt-live-1" The Live model. Required in the session configuration for every transport; do not pass it as a URL query parameter.
One of the following: string minLength1 "gpt-live-1" The Live model. Required in the session configuration for every transport; do not pass it as a URL query parameter.
audio: optional { format, output } Startup audio configuration. Only primary WebSockets accept audio.format; WebRTC and SIP negotiate their media format. Voice and format are immutable after startup.
format: optional AudioFormat Audio encoding and sample rate for audio sent and received over a Live WebSocket connection. WebRTC and SIP negotiate their media format separately.
output: optional { voice } The voice used for speech generated by the Live model.
voice: optional string or "alloy" or "ash" or "ballad" or 35 more or CustomVoice The voice used for Live speech, as a built-in voice name or a custom voice object containing its ID. Defaults to marin and cannot change after startup.
One of the following: string "alloy" or "ash" or "ballad" or 35 more The voice used for Live speech, as a built-in voice name or a custom voice object containing its ID. Defaults to marin and cannot change after startup.
One of the following: "alloy" "ash" "ballad" "beacon" "bossa" "brise" "calandria" "cedar" "cinder" "compas" "coral" "delta" "destello" "echo" "flitz" "gleam" "halo" "harema" "hornero" "juni" "luciernaga" "marin" "meridian" "nira" "noeul" "nuri" "pearl" "quartz" "ripple" "sage" "shida" "shimmer" "sillage" "stone" "tempo" "verse" "vesper" "willow" CustomVoice { id } id: string minLength1 maxLength128 client: optional ClientConfig { data_channel } Startup-only capabilities for an untrusted frontend attached to a unified WebRTC session. Trusted sideband connections are unaffected.
delegation: optional ClientDelegation or Responses or Decisions or null Who handles tasks delegated by the Live model. Omitted or null selects your application; use responses or decisions to delegate to the corresponding API.
One of the following: ClientDelegation { type, acknowledgement_delay_ms } Delegate tasks to your application. The Live session emits delegation events that your backend handles.
type: "client" (default: "client") The delegation owner. Always client for tasks handled by your application.
acknowledgement_delay_ms: optional integer Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations.
minimum0 maximum800 Responses { responses, type, acknowledgement_delay_ms } Delegate tasks to a Responses model managed by the Live session.
responses: ResponsesDelegationConfig { model, instructions, max_output_tokens, 6 more } Backend model, prompt, and tools used when the Live session delegates a task to Responses.
type: "responses" (default: "responses") The delegation owner. Always responses for tasks handled by the Responses API.
acknowledgement_delay_ms: optional integer Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations.
minimum0 maximum800 Decisions { decisions, type, acknowledgement_delay_ms } decisions: DecisionsDelegationConfig { model, questions } type: "decisions" (default: "decisions") acknowledgement_delay_ms: optional integer Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400. Updates apply to subsequent delegations.
minimum0 maximum800 input: optional array of InitialItem Ordered text-only history supplied before startup. Supports developer, user, and assistant messages with one text part each; at most 128 messages and a configured rendered-token limit (8,192 by default).
maxItems128 One of the following: Developer { content, role, id, 2 more } A developer message included in the initial text history of a Live session.
content: array of { text, type } The message content. Supply exactly one text part for the initial Live conversation history.
minItems1 maxItems1 text: string The message text to include in the Live session’s initial conversation history.
type: optional "input_text" (default: "input_text") The text content type. Always input_text.
role: "developer" (default: "developer") The author of this history message. Always developer.
id: optional string or null An optional identifier for the supplied history message. Live uses the message’s role and text to initialize the conversation.
status: optional "incomplete" or "completed" or null The supplied message’s status. Live uses its text as history and does not resume an incomplete message.
One of the following: "incomplete" "completed" type: optional "message" (default: "message") The history item type. Always message.
User { content, role, id, 2 more } A user message included in the initial text history of a Live session.
content: array of { text, type } The message content. Supply exactly one text part for the initial Live conversation history.
minItems1 maxItems1 text: string The message text to include in the Live session’s initial conversation history.
type: optional "input_text" (default: "input_text") The text content type. Always input_text.
role: "user" (default: "user") The author of this history message. Always user.
id: optional string or null An optional identifier for the supplied history message. Live uses the message’s role and text to initialize the conversation.
status: optional "incomplete" or "completed" or null The supplied message’s status. Live uses its text as history and does not resume an incomplete message.
One of the following: "incomplete" "completed" type: optional "message" (default: "message") The history item type. Always message.
Assistant { content, role, id, 2 more } An assistant message included in the initial text history of a Live session.
content: array of Text or OutputText The message content. Supply exactly one text part for the initial Live conversation history.
minItems1 maxItems1 One of the following: Text { text, type } Assistant text supplied as conversation history when starting a Live session.
text: string The message text to include in the Live session’s initial conversation history.
type: optional "text" (default: "text") The text content type. Always text.
OutputText { text, type } Assistant output text supplied as conversation history when starting a Live session.
text: string The message text to include in the Live session’s initial conversation history.
type: "output_text" (default: "output_text") The text content type. Always output_text.
role: "assistant" (default: "assistant") The author of this history message. Always assistant.
id: optional string or null An optional identifier for the supplied history message. Live uses the message’s role and text to initialize the conversation.
status: optional "incomplete" or "completed" or null The supplied message’s status. Live uses its text as history and does not resume an incomplete message.
One of the following: "incomplete" "completed" type: optional "message" (default: "message") The history item type. Always message.
instructions: optional string or null Frontend instructions for voice, conversation, interruptions, and when to delegate. Start with the Live prompting guide; put business rules and tool workflows in a separate backend prompt. Limited to 16,384 client-supplied tokens. Omitted or blank instructions use server defaults. Immutable after startup.
safety_identifier: optional string or null A stable identifier for your end user, used for safety monitoring and abuse detection.
maxLength128 store: optional boolean Whether to store the session for later forking and recording download. Defaults to false for new sessions.
triggers: optional array of DecisionTrigger { decisions, event_type, type } At most one Decisions trigger per transcript event type. Triggers run independently of delegation. On update, omit triggers to keep the current configuration, send a list to replace it, or send an empty list to disable triggers.
maxItems2 decisions: { questions } Questions evaluated using the current transcript and the session’s shared Decision items.
questions: array of DecisionQuestion Native Decision API questions evaluated together, in order.
minItems1 maxItems64 One of the following: Choice { choices, instructions, type, name } choices: array of DecisionChoiceOption { value, description } minItems2 maxItems255 value: string or boolean One of the following: string boolean description: optional string instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "choice" (default: "choice") name: optional string Predicate { instructions, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "predicate" (default: "predicate") name: optional string Score { instructions, levels, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") levels: array of { label, description } minItems2 maxItems10 label: string description: optional string type: "score" (default: "score") name: optional string event_type: "session.input_transcript.delta" or "session.output_transcript.delta" One of the following: "session.input_transcript.delta" "session.output_transcript.delta" type: "decisions" (default: "decisions")
SessionResource object { id, expires_at, model, 9 more } The resolved Live session configuration and server-assigned session metadata.
SessionResource object { id, expires_at, model, 9 more }
The resolved Live session configuration and server-assigned session metadata.
id: string The unique ID of the Live session. Use this ID for sideband connections, forking, and recording download.
expires_at: integer The Unix timestamp, in seconds, at which the Live session expires.
model: string or "gpt-live-1" The Live model. Required in the session configuration for every transport; do not pass it as a URL query parameter.
One of the following: string minLength1 "gpt-live-1" The Live model. Required in the session configuration for every transport; do not pass it as a URL query parameter.
status: "active" (default: "active") The status of the session snapshot. Always active, including the final snapshot in session.closed; use the event type to determine that the session has closed.
audio: optional { format, output } Startup audio configuration. Only primary WebSockets accept audio.format; WebRTC and SIP negotiate their media format. Voice and format are immutable after startup.
format: optional AudioFormat Audio encoding and sample rate for audio sent and received over a Live WebSocket connection. WebRTC and SIP negotiate their media format separately.
output: optional { voice } The voice used for speech generated by the Live model.
voice: optional string or "alloy" or "ash" or "ballad" or 35 more or CustomVoice The voice used for Live speech, as a built-in voice name or a custom voice object containing its ID. Defaults to marin and cannot change after startup.
One of the following: string "alloy" or "ash" or "ballad" or 35 more The voice used for Live speech, as a built-in voice name or a custom voice object containing its ID. Defaults to marin and cannot change after startup.
One of the following: "alloy" "ash" "ballad" "beacon" "bossa" "brise" "calandria" "cedar" "cinder" "compas" "coral" "delta" "destello" "echo" "flitz" "gleam" "halo" "harema" "hornero" "juni" "luciernaga" "marin" "meridian" "nira" "noeul" "nuri" "pearl" "quartz" "ripple" "sage" "shida" "shimmer" "sillage" "stone" "tempo" "verse" "vesper" "willow" CustomVoice { id } id: string minLength1 maxLength128 client: optional ClientConfig { data_channel } Startup-only capabilities for an untrusted frontend attached to a unified WebRTC session. Trusted sideband connections are unaffected.
delegation: optional ClientDelegation or Responses or Decisions or null Who handles tasks delegated by the Live model. Omitted or null selects your application; use responses or decisions to delegate to the corresponding API.
One of the following: ClientDelegation { type, acknowledgement_delay_ms } Delegate tasks to your application. The Live session emits delegation events that your backend handles.
type: "client" (default: "client") The delegation owner. Always client for tasks handled by your application.
acknowledgement_delay_ms: optional integer Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations.
minimum0 maximum800 Responses { responses, type, acknowledgement_delay_ms } Delegate tasks to a Responses model managed by the Live session.
responses: ResponsesDelegationConfig { model, instructions, max_output_tokens, 6 more } Backend model, prompt, and tools used when the Live session delegates a task to Responses.
type: "responses" (default: "responses") The delegation owner. Always responses for tasks handled by the Responses API.
acknowledgement_delay_ms: optional integer Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations.
minimum0 maximum800 Decisions { decisions, type, acknowledgement_delay_ms } decisions: DecisionsDelegationConfig { model, questions } type: "decisions" (default: "decisions") acknowledgement_delay_ms: optional integer Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400. Updates apply to subsequent delegations.
minimum0 maximum800 input: optional array of InitialItem Ordered text-only history supplied before startup. Supports developer, user, and assistant messages with one text part each; at most 128 messages and a configured rendered-token limit (8,192 by default).
maxItems128 One of the following: Developer { content, role, id, 2 more } A developer message included in the initial text history of a Live session.
content: array of { text, type } The message content. Supply exactly one text part for the initial Live conversation history.
minItems1 maxItems1 text: string The message text to include in the Live session’s initial conversation history.
type: optional "input_text" (default: "input_text") The text content type. Always input_text.
role: "developer" (default: "developer") The author of this history message. Always developer.
id: optional string or null An optional identifier for the supplied history message. Live uses the message’s role and text to initialize the conversation.
status: optional "incomplete" or "completed" or null The supplied message’s status. Live uses its text as history and does not resume an incomplete message.
One of the following: "incomplete" "completed" type: optional "message" (default: "message") The history item type. Always message.
User { content, role, id, 2 more } A user message included in the initial text history of a Live session.
content: array of { text, type } The message content. Supply exactly one text part for the initial Live conversation history.
minItems1 maxItems1 text: string The message text to include in the Live session’s initial conversation history.
type: optional "input_text" (default: "input_text") The text content type. Always input_text.
role: "user" (default: "user") The author of this history message. Always user.
id: optional string or null An optional identifier for the supplied history message. Live uses the message’s role and text to initialize the conversation.
status: optional "incomplete" or "completed" or null The supplied message’s status. Live uses its text as history and does not resume an incomplete message.
One of the following: "incomplete" "completed" type: optional "message" (default: "message") The history item type. Always message.
Assistant { content, role, id, 2 more } An assistant message included in the initial text history of a Live session.
content: array of Text or OutputText The message content. Supply exactly one text part for the initial Live conversation history.
minItems1 maxItems1 One of the following: Text { text, type } Assistant text supplied as conversation history when starting a Live session.
text: string The message text to include in the Live session’s initial conversation history.
type: optional "text" (default: "text") The text content type. Always text.
OutputText { text, type } Assistant output text supplied as conversation history when starting a Live session.
text: string The message text to include in the Live session’s initial conversation history.
type: "output_text" (default: "output_text") The text content type. Always output_text.
role: "assistant" (default: "assistant") The author of this history message. Always assistant.
id: optional string or null An optional identifier for the supplied history message. Live uses the message’s role and text to initialize the conversation.
status: optional "incomplete" or "completed" or null The supplied message’s status. Live uses its text as history and does not resume an incomplete message.
One of the following: "incomplete" "completed" type: optional "message" (default: "message") The history item type. Always message.
instructions: optional string or null Frontend instructions for voice, conversation, interruptions, and when to delegate. Start with the Live prompting guide; put business rules and tool workflows in a separate backend prompt. Limited to 16,384 client-supplied tokens. Omitted or blank instructions use server defaults. Immutable after startup.
safety_identifier: optional string or null A stable identifier for your end user, used for safety monitoring and abuse detection.
maxLength128 store: optional boolean Whether to store the session for later forking and recording download. Defaults to false for new sessions.
triggers: optional array of DecisionTrigger { decisions, event_type, type } At most one Decisions trigger per transcript event type. Triggers run independently of delegation. On update, omit triggers to keep the current configuration, send a list to replace it, or send an empty list to disable triggers.
maxItems2 decisions: { questions } Questions evaluated using the current transcript and the session’s shared Decision items.
questions: array of DecisionQuestion Native Decision API questions evaluated together, in order.
minItems1 maxItems64 One of the following: Choice { choices, instructions, type, name } choices: array of DecisionChoiceOption { value, description } minItems2 maxItems255 value: string or boolean One of the following: string boolean description: optional string instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "choice" (default: "choice") name: optional string Predicate { instructions, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "predicate" (default: "predicate") name: optional string Score { instructions, levels, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") levels: array of { label, description } minItems2 maxItems10 label: string description: optional string type: "score" (default: "score") name: optional string event_type: "session.input_transcript.delta" or "session.output_transcript.delta" One of the following: "session.input_transcript.delta" "session.output_transcript.delta" type: "decisions" (default: "decisions")
SessionStartEvent object { session, type, event_id } Start a Live session on a primary WebSocket. Send this event before other commands and wait for session.started.
SessionStartEvent object { session, type, event_id }
Start a Live session on a primary WebSocket. Send this event before other commands and wait for session.started.
session: SessionConfig { model, audio, client, 6 more } Initial configuration for a primary WebSocket. Send session.start first and wait for session.started before application commands. WebRTC creation already starts the session; do not send this event again on its data channel.
type: "session.start" (default: "session.start") The Live client event type. Always session.start.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512
SessionStartedEvent object { event_id, session, type, client_event_id } Returned when a Live session has started. Contains the resolved session configuration, including server defaults.
SessionStartedEvent object { event_id, session, type, client_event_id }
Returned when a Live session has started. Contains the resolved session configuration, including server defaults.
event_id: string The unique ID of the Live server event.
session: SessionResource { id, expires_at, model, 9 more } The resolved Live session configuration and server-assigned session metadata.
type: "session.started" (default: "session.started") The event type, always session.started.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
SessionUpdateConfig object { delegation, triggers } Changes to an active Live session’s delegation settings and transcript triggers.
SessionUpdateConfig object { delegation, triggers }
Changes to an active Live session’s delegation settings and transcript triggers.
delegation: optional ClientDelegation or Responses or Decisions or null Delegation settings to update. The delegation type must match the current session; omitted settings retain their values.
One of the following: ClientDelegation { type, acknowledgement_delay_ms } Delegate tasks to your application. The Live session emits delegation events that your backend handles.
type: "client" (default: "client") The delegation owner. Always client for tasks handled by your application.
acknowledgement_delay_ms: optional integer Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations.
minimum0 maximum800 Responses { type, acknowledgement_delay_ms, responses } Update the Responses backend for an existing Live session without changing delegation ownership.
type: "responses" (default: "responses") The delegation owner. Always responses for tasks handled by the Responses API.
acknowledgement_delay_ms: optional integer Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations.
minimum0 maximum800 responses: optional ResponsesDelegationUpdateConfig { instructions, max_output_tokens, model, 6 more } Responses backend settings to update. Omitted settings keep their existing values.
Decisions { type, acknowledgement_delay_ms, decisions } type: "decisions" (default: "decisions") acknowledgement_delay_ms: optional integer Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations.
minimum0 maximum800 decisions: optional DecisionsDelegationUpdateConfig { model, questions } triggers: optional array of DecisionTrigger { decisions, event_type, type } At most one Decisions trigger per transcript event type. Triggers run independently of delegation. On update, omit triggers to keep the current configuration, send a list to replace it, or send an empty list to disable triggers.
maxItems2 decisions: { questions } Questions evaluated using the current transcript and the session’s shared Decision items.
questions: array of DecisionQuestion Native Decision API questions evaluated together, in order.
minItems1 maxItems64 One of the following: Choice { choices, instructions, type, name } choices: array of DecisionChoiceOption { value, description } minItems2 maxItems255 value: string or boolean One of the following: string boolean description: optional string instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "choice" (default: "choice") name: optional string Predicate { instructions, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") type: "predicate" (default: "predicate") name: optional string Score { instructions, levels, type, name } instructions: string or array of DecisionInputText or InputImage or InputJSON One of the following: string array of DecisionInputText or InputImage or InputJSON minItems1 One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") levels: array of { label, description } minItems2 maxItems10 label: string description: optional string type: "score" (default: "score") name: optional string event_type: "session.input_transcript.delta" or "session.output_transcript.delta" One of the following: "session.input_transcript.delta" "session.output_transcript.delta" type: "decisions" (default: "decisions")
SessionUpdateEvent object { session, type, event_id } Update the delegation settings or transcript triggers of an active Live session. The server acknowledges accepted changes with session.updated.
SessionUpdateEvent object { session, type, event_id }
Update the delegation settings or transcript triggers of an active Live session. The server acknowledges accepted changes with session.updated.
session: SessionUpdateConfig { delegation, triggers } Sparse delegation and trigger updates. Omitted settings retain their values. The delegation type cannot change. Model, frontend instructions, audio, and startup input are immutable.
type: "session.update" (default: "session.update") The Live client event type. Always session.update.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512
SessionUpdatedEvent object { event_id, session, type, client_event_id } Returned when a Live session update is accepted. Contains the resolved session configuration after the update.
SessionUpdatedEvent object { event_id, session, type, client_event_id }
Returned when a Live session update is accepted. Contains the resolved session configuration after the update.
event_id: string The unique ID of the Live server event.
session: SessionResource { id, expires_at, model, 9 more } The resolved Live session configuration and server-assigned session metadata.
type: "session.updated" (default: "session.updated") The event type, always session.updated.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
SessionUsageUpdatedEvent object { event_id, type, usage, 2 more } Reports cumulative Live audio usage and, when available, the most recent context-window usage. Delegated Responses token usage is reported separately in response.event events.
SessionUsageUpdatedEvent object { event_id, type, usage, 2 more }
Reports cumulative Live audio usage and, when available, the most recent context-window usage. Delegated Responses token usage is reported separately in response.event events.
event_id: string The unique ID of the Live server event.
type: "session.usage.updated" (default: "session.usage.updated") The event type, always session.usage.updated.
usage: SessionUsage { seconds } The cumulative Live audio usage so far.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
context_window: optional { usage_ratio } The latest measured Live context-window usage. Omitted when the context limit is unknown.
usage_ratio: number The latest active context token count divided by the Live model context limit. Can decrease after compaction and may lag between measured audio frames.
ThinkingAppendEvent object { content, delegation_id, type, event_id } Provide silent reasoning or progress context to the Live model, optionally for an existing client delegation. Rejected while an audio fixture is active.
ThinkingAppendEvent object { content, delegation_id, type, event_id }
Provide silent reasoning or progress context to the Live model, optionally for an existing client delegation. Rejected while an audio fixture is active.
content: string Silent reasoning or progress context, limited to 500 tokens. It does not directly request speech, but can influence later speech and is not a secrecy boundary.
delegation_id: string or null Required, nullable. Set null for general session context, or use the ID from session.delegation.created for an existing client delegation. Non-null IDs are not accepted with Responses delegation.
minLength1 type: "session.thinking.append" (default: "session.thinking.append") The Live client event type. Always session.thinking.append.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512
ThinkingAppendedEvent object { end_ms, event_id, start_ms, 2 more } Returned when a session.thinking.append command is accepted into the Live session timeline. Acknowledges the added reasoning context without guaranteeing any spoken output.
ThinkingAppendedEvent object { end_ms, event_id, start_ms, 2 more }
Returned when a session.thinking.append command is accepted into the Live session timeline. Acknowledges the added reasoning context without guaranteeing any spoken output.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.thinking.appended" (default: "session.thinking.appended") The event type, always session.thinking.appended.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
LiveForks
Connect
WebSocket
/live/sessions/{session_id}/forkModels
ForkClientEvent = ForkSessionStartEvent { session, type, event_id } or SessionUpdateEvent { session, type, event_id } or InputAudioAppendEvent { audio, type, event_id } or 12 moreClient events for a Live fork WebSocket. First send session.start with an overrides object (which may be empty), then wait for session.started before sending other commands. The model and conversation are inherited from the stored session.
ForkClientEvent = ForkSessionStartEvent { session, type, event_id } or SessionUpdateEvent { session, type, event_id } or InputAudioAppendEvent { audio, type, event_id } or 12 more
Client events for a Live fork WebSocket. First send session.start with an overrides object (which may be empty), then wait for session.started before sending other commands. The model and conversation are inherited from the stored session.
One of the following: ForkSessionStartEvent { session, type, event_id } Start a Live session after connecting to a stored session’s fork WebSocket. Send an empty session object to use the stored configuration.
session: ForkSessionConfig { audio, client, delegation, store } Overrides for a stored session after connecting to the fork WebSocket. An empty object inherits the stored configuration; do not supply a new model. audio.format applies only to the new WebSocket connection. client overrides are only supported for WebRTC forks.
type: "session.start" (default: "session.start") The Live client event type. Always session.start.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 SessionUpdateEvent { session, type, event_id } Update the delegation settings or transcript triggers of an active Live session. The server acknowledges accepted changes with session.updated.
session: SessionUpdateConfig { delegation, triggers } Sparse delegation and trigger updates. Omitted settings retain their values. The delegation type cannot change. Model, frontend instructions, audio, and startup input are immutable.
type: "session.update" (default: "session.update") The Live client event type. Always session.update.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 InputAudioAppendEvent { audio, type, event_id } Send audio to a Live session over its primary WebSocket. WebRTC and SIP sessions send audio over their media transport.
audio: string Base64-encoded raw audio in the startup-selected format, without a WAV or other container header. Primary WebSocket only; media transports use their audio track. Audio appends have no acknowledgment. Reflected sideband server events reuse this event type and audio key, with no timestamps or event_id; their audio is always mono PCM16LE at 24 kHz.
minLength1 type: "session.input_audio.append" (default: "session.input_audio.append") The Live client event type. Always session.input_audio.append.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 InputAudioMuteEvent { type, event_id } Mute audio input to the Live model without closing the session. The server acknowledges with session.input_audio.muted.
type: "session.input_audio.mute" (default: "session.input_audio.mute") The Live client event type. Always session.input_audio.mute.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 InputAudioUnmuteEvent { type, event_id } Resume audio input to a Live model after muting it. The server acknowledges with session.input_audio.unmuted. An active audio fixture still pauses inference until playback finishes.
type: "session.input_audio.unmute" (default: "session.input_audio.unmute") The Live client event type. Always session.input_audio.unmute.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 OutputAudioFixturePlayEvent { audio, type, event_id } Play one complete audio fixture through the session output. Send this event over the primary WebSocket or a server sideband. Audio must be raw mono PCM16 little-endian at 24 kHz, regardless of the session audio format, and at most 30 seconds. The entire payload is validated before model speech is interrupted and input is muted. Only one fixture can be active. fixture.playing acknowledges acceptance and fixture.played reports server mixer drain. These notifications are not exact transport-media boundaries. Input arriving during playback is dropped and not replayed.
audio: string Base64-encoded raw mono PCM16 audio, little-endian, at 24 kHz. Do not include a WAV or other container header.
minLength1 maxLength1920000 type: "session.output_audio.fixture.play" (default: "session.output_audio.fixture.play") event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 InstructionsAppendEvent { content, delegation_id, type, event_id } Append developer instructions to the Live conversation while it is running, optionally associating them with an existing client delegation. Rejected while an audio fixture is active. These instructions remain in conversation history through compaction and are included in future backend delegations. For external audio, describe content already delivered so the assistant does not repeat it.
content: string Instruction text to append, limited to 500 tokens. This is a plain string, not an array of content parts.
delegation_id: string or null Required, nullable. Set null for general session context, or use the ID from session.delegation.created for an existing client delegation. Non-null IDs are not accepted with Responses delegation.
minLength1 type: "session.instructions.append" (default: "session.instructions.append") The Live client event type. Always session.instructions.append.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 ThinkingAppendEvent { content, delegation_id, type, event_id } Provide silent reasoning or progress context to the Live model, optionally for an existing client delegation. Rejected while an audio fixture is active.
content: string Silent reasoning or progress context, limited to 500 tokens. It does not directly request speech, but can influence later speech and is not a secrecy boundary.
delegation_id: string or null Required, nullable. Set null for general session context, or use the ID from session.delegation.created for an existing client delegation. Non-null IDs are not accepted with Responses delegation.
minLength1 type: "session.thinking.append" (default: "session.thinking.append") The Live client event type. Always session.thinking.append.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 CommentaryAppendEvent { content, delegation_id, type, event_id } Provide context the Live model can communicate to the user, optionally for an existing client delegation. Rejected while an audio fixture is active.
content: string Speakable context for the Live model, limited to 500 tokens. Use this for a result the model should communicate; use session.thinking.append for silent context.
delegation_id: string or null Required, nullable. Set null for general session context, or use the ID from session.delegation.created for an existing client delegation. Non-null IDs are not accepted with Responses delegation.
minLength1 type: "session.commentary.append" (default: "session.commentary.append") The Live client event type. Always session.commentary.append.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 DecisionCreateEvent { event_id, type } Evaluate the configured questions using the current conversation and Decision inputs. Reusing an event_id is rejected for the lifetime of the session.
event_id: string minLength1 maxLength256 type: "decision.create" (default: "decision.create") DecisionItemCreateEvent { item, type, event_id } Append an input item for future decisions without starting an evaluation or changing Live voice context.
item: Message or Message or Message or 5 more One of the following: Message { content, role, id, type } content: string or array of InputText or InputImage or InputJSON The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of InputText or InputImage or InputJSON One of the following: InputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") role: "user" (default: "user") The message role. Always user.
id: optional string or null The unique ID of this message item.
type: optional "message" (default: "message") Message { content, role, id, type } content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "developer" (default: "developer") id: optional string or null The unique ID of this message item.
type: optional "message" (default: "message") Message { content, role, id, type } content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "system" (default: "system") id: optional string or null The unique ID of this message item.
type: optional "message" (default: "message") Message { content, role, id, 2 more } content: string or array of OutputText or Refusal The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of OutputText or Refusal One of the following: OutputText { text, type } text: string The text content.
maxLength10485760 type: "output_text" (default: "output_text") Refusal { refusal, type } refusal: string The refusal text.
maxLength10485760 type: "refusal" (default: "refusal") role: "assistant" (default: "assistant") id: optional string or null The unique ID of this message item.
phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer). For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" type: optional "message" (default: "message") FunctionCall { arguments, call_id, name, 4 more } arguments: string The function arguments as a JSON string.
call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 name: string pattern^[a-zA-Z0-9_.-]+$ minLength1 maxLength128 type: "function_call" (default: "function_call") The item type. Always function_call.
id: optional string or null The local Decision item ID.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" FunctionCallOutput { call_id, output, type, 2 more } call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "function_call_output" (default: "function_call_output") The type of the function tool call output. Always function_call_output.
id: optional string or null The local Decision item ID.
status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCall { call_id, input, name, 4 more } call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 input: string The input to the custom tool call.
name: string The name of the custom tool to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 type: "custom_tool_call" (default: "custom_tool_call") The item type. Always custom_tool_call.
id: optional string or null The local Decision item ID.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCallOutput { call_id, output, type, 2 more } call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "custom_tool_call_output" (default: "custom_tool_call_output") id: optional string or null The local Decision item ID.
status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" type: "decision.item.create" (default: "decision.item.create") event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 DecisionItemDeleteEvent { item_id, type, event_id } Delete an input item from future decisions without starting an evaluation or changing Live voice context.
item_id: string minLength1 type: "decision.item.delete" (default: "decision.item.delete") event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 ResponseItemCreateEvent { item, type, event_id } Add an input item to the Live session’s Responses backend. Requires Responses delegation; use response.create to request a response.
item: EasyInputMessage or Message or ResponseOutputMessage or 30 more An input item to append to the Responses backend conversation, such as a user message or a function tool result.
One of the following: EasyInputMessage { content, role, phase, type } A message input to the model with a role indicating instruction following
hierarchy. Instructions given with the developer or system role take
precedence over instructions given with the user role. Messages with the
assistant role are presumed to have been generated by the model in previous
interactions.
content: string or ResponseInputMessageContentList Text, image, or audio input to the model, used to generate a response.
Can also contain previous assistant responses.
One of the following: TextInput = string A text input to the model.
ResponseInputMessageContentList = array of ResponseInputContent A list of one or many input items to the model, containing different content
types.
One of the following: ResponseInputText { text, type, prompt_cache_breakpoint } A text input to the model.
text: string The text input to the model.
type: "input_text" (default: "input_text") The type of the input item. Always input_text.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputImage { detail, type, file_id, 2 more } An image input to the model. Learn about image inputs.
detail: ImageDetail The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
type: "input_image" (default: "input_image") The type of the input item. Always input_image.
file_id: optional string or null The ID of the file to be sent to the model.
image_url: optional string or null The URL of the image to be sent to the model. A fully qualified URL or base64 encoded image in a data URL.
formaturi prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputFile { type, detail, file_data, 4 more } A file input to the model.
type: "input_file" (default: "input_file") The type of the input item. Always input_file.
detail: optional "auto" or "low" or "high" The detail level of the file to be sent to the model. Use auto to let the system select the detail level; for GPT-5.6 and later models, auto uses high-quality rendering, which may increase input token usage. Use low for lower-cost rendering, or high to render the file at higher quality. Defaults to auto.
One of the following: "auto" "low" "high" file_data: optional string The content of the file to be sent to the model.
file_id: optional string or null The ID of the file to be sent to the model.
file_url: optional string The URL of the file to be sent to the model.
formaturi filename: optional string The name of the file to be sent to the model.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" role: "user" or "assistant" or "system" or "developer" The role of the message input. One of user, assistant, system, or
developer.
One of the following: "user" "assistant" "system" "developer" phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer).
For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend
phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" type: optional "message" The type of the message input. Always message.
Message { content, role, status, type } A message input to the model with a role indicating instruction following
hierarchy. Instructions given with the developer or system role take
precedence over instructions given with the user role.
content: ResponseInputMessageContentList { , , } A list of one or many input items to the model, containing different content
types.
role: "user" or "system" or "developer" The role of the message input. One of user, system, or developer.
One of the following: "user" "system" "developer" status: optional "in_progress" or "completed" or "incomplete" The status of item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" type: optional "message" The type of the message input. Always set to message.
ResponseOutputMessage { id, content, role, 3 more } An output message from the model.
id: string The unique ID of the output message.
content: array of ResponseOutputText or ResponseOutputRefusal The content of the output message.
One of the following: ResponseOutputText { annotations, logprobs, text, type } A text output from the model.
annotations: array of FileCitation or URLCitation or ContainerFileCitation or FilePath The annotations of the text output.
One of the following: FileCitation { file_id, filename, index, type } A citation to a file.
file_id: string The ID of the file.
filename: string The filename of the file cited.
index: integer The index in the output text at which to insert the file citation.
type: "file_citation" (default: "file_citation") The type of the file citation. Always file_citation.
URLCitation { end_index, start_index, title, 2 more } A citation for a web resource used to generate a model response.
end_index: integer The index of the last character of the URL citation in the message.
start_index: integer The index of the first character of the URL citation in the message.
title: string The title of the web resource.
type: "url_citation" (default: "url_citation") The type of the URL citation. Always url_citation.
url: string The URL of the web resource.
formaturi ContainerFileCitation { container_id, end_index, file_id, 3 more } A citation for a container file used to generate a model response.
container_id: string The ID of the container file.
end_index: integer The index of the last character of the container file citation in the message.
file_id: string The ID of the file.
filename: string The filename of the container file cited.
start_index: integer The index of the first character of the container file citation in the message.
type: "container_file_citation" (default: "container_file_citation") The type of the container file citation. Always container_file_citation.
FilePath { file_id, index, type } A path to a file.
file_id: string The ID of the file.
index: integer The index of the file in the list of files.
type: "file_path" The type of the file path. Always file_path.
logprobs: array of { token, bytes, logprob, top_logprobs } token: string bytes: array of number logprob: number top_logprobs: array of { token, bytes, logprob } token: string bytes: array of number logprob: number text: string The text output from the model.
type: "output_text" (default: "output_text") The type of the output text. Always output_text.
ResponseOutputRefusal { refusal, type } A refusal from the model.
refusal: string The refusal explanation from the model.
type: "refusal" (default: "refusal") The type of the refusal. Always refusal.
role: "assistant" The role of the output message. Always assistant.
status: "in_progress" or "completed" or "incomplete" The status of the message input. One of in_progress, completed, or
incomplete. Populated when input items are returned via API.
One of the following: "in_progress" "completed" "incomplete" type: "message" The type of the output message. Always message.
phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer).
For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend
phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" FileSearchCall { id, queries, status, 2 more } The results of a file search tool call. See the
file search guide for more information.
id: string The unique ID of the file search tool call.
queries: array of string The queries used to search for files.
status: "in_progress" or "searching" or "completed" or 2 more The status of the file search tool call. One of in_progress,
searching, incomplete or failed,
One of the following: "in_progress" "searching" "completed" "incomplete" "failed" type: "file_search_call" The type of the file search tool call. Always file_search_call.
results: optional array of object or null The results of the file search tool call.
attributes: optional map[string or number or boolean] or null Set of 16 key-value pairs that can be attached to an object. This can be
useful for storing additional information about the object in a structured
format, and querying for objects via API or the dashboard. Keys are strings
with a maximum length of 64 characters. Values are strings with a maximum
length of 512 characters, booleans, or numbers.
One of the following: string maxLength512 number boolean file_id: optional string The unique ID of the file.
filename: optional string The name of the file.
score: optional number The relevance score of the file - a value between 0 and 1.
formatfloat text: optional string The text that was retrieved from the file.
ComputerCall { id, call_id, pending_safety_checks, 4 more } A tool call to a computer use tool. See the
computer use guide for more information.
id: string The unique ID of the computer call.
call_id: string An identifier used when responding to the tool call with output.
pending_safety_checks: array of { id, code, message } The pending safety checks for the computer call.
id: string The ID of the pending safety check.
code: optional string or null The type of the pending safety check.
message: optional string or null Details about the pending safety check.
status: "in_progress" or "completed" or "incomplete" The status of the item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" type: "computer_call" (default: "computer_call") The type of the computer call. Always computer_call.
action: optional ComputerAction A click action.
actions: optional ComputerActionList { Click, DoubleClick, Drag, 6 more } Flattened batched actions for computer_use. Each action includes an
type discriminator and action-specific fields.
ComputerCallOutput { call_id, output, type, 3 more } The output of a computer tool call.
call_id: string The ID of the computer tool call that produced the output.
minLength1 maxLength64 output: ResponseComputerToolCallOutputScreenshot { type, file_id, image_url } A computer screenshot image used with the computer use tool.
type: "computer_call_output" (default: "computer_call_output") The type of the computer tool call output. Always computer_call_output.
id: optional string or null The ID of the computer tool call output.
acknowledged_safety_checks: optional array of object or null The safety checks reported by the API that have been acknowledged by the developer.
id: string The ID of the pending safety check.
code: optional string or null The type of the pending safety check.
message: optional string or null Details about the pending safety check.
status: optional "in_progress" or "completed" or "incomplete" or null The status of the message input. One of in_progress, completed, or incomplete. Populated when input items are returned via API.
One of the following: "in_progress" "completed" "incomplete" WebSearchCall { id, status, type, action } The results of a web search tool call. See the
web search guide for more information.
id: string The unique ID of the web search tool call.
status: "in_progress" or "searching" or "completed" or 2 more The status of the web search tool call.
One of the following: "in_progress" "searching" "completed" "failed" "incomplete" type: "web_search_call" The type of the web search tool call. Always web_search_call.
action: optional Search or OpenPage or FindInPage An object describing the specific action taken in this web search call.
Includes details on how the model used the web (search, open_page, find_in_page).
One of the following: Search { type, queries, query, sources } Action type “search” - Performs a web search query.
type: "search" The action type.
queries: optional array of string The search queries.
query: optional string Deprecated.
The search query.
sources: optional array of { type, url } The sources used in the search.
type: "url" The type of source. Always url.
url: string The URL of the source.
formaturi OpenPage { type, url } Action type “open_page” - Opens a specific URL from search results.
type: "open_page" The action type.
url: optional string or null The URL opened by the model.
formaturi FindInPage { pattern, type, url } Action type “find_in_page”: Searches for a pattern within a loaded page.
pattern: string The pattern or text to search for within the page.
type: "find_in_page" The action type.
url: optional string or null The URL of the page searched for the pattern, when available.
formaturi FunctionCall { arguments, call_id, name, 6 more } A tool call to run a function. See the
function calling guide for more information.
arguments: string A JSON string of the arguments to pass to the function.
call_id: string The unique ID of the function tool call generated by the model.
name: string The name of the function to run.
type: "function_call" The type of the function tool call. Always function_call.
id: optional string The unique ID of the function tool call.
async: optional boolean Whether the function tool call runs asynchronously.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
type: "program" (default: "program") namespace: optional string The namespace of the function to run.
status: optional "in_progress" or "completed" or "incomplete" The status of the item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" FunctionCallOutput { output, type, id, 5 more } The output of a function tool call.
output: string or array of ResponseInputTextContent or ResponseInputImageContent or ResponseInputFileContent Text, image, or file output of the function tool call.
One of the following: string A JSON string of the output of the function tool call.
maxLength10485760 array of ResponseInputTextContent or ResponseInputImageContent or ResponseInputFileContent An array of content outputs (text, image, file) for the function tool call.
One of the following: ResponseInputTextContent { text, type, prompt_cache_breakpoint } A text input to the model.
text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") The type of the input item. Always input_text.
prompt_cache_breakpoint: optional object or null Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputImageContent { type, detail, file_id, 2 more } An image input to the model. Learn about image inputs
type: "input_image" (default: "input_image") The type of the input item. Always input_image.
detail: optional ImageDetail or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
file_id: optional string or null The ID of the file to be sent to the model.
image_url: optional string or null The URL of the image to be sent to the model. A fully qualified URL or base64 encoded image in a data URL.
maxLength20971520 formaturi prompt_cache_breakpoint: optional object or null Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputFileContent { type, detail, file_data, 4 more } A file input to the model.
type: "input_file" (default: "input_file") The type of the input item. Always input_file.
detail: optional "auto" or "low" or "high" The detail level of the file to be sent to the model. Use auto to let the system select the detail level; for GPT-5.6 and later models, auto uses high-quality rendering, which may increase input token usage. Use low for lower-cost rendering, or high to render the file at higher quality. Defaults to auto.
One of the following: "auto" "low" "high" file_data: optional string or null The base64-encoded data of the file to be sent to the model.
maxLength73400320 file_id: optional string or null The ID of the file to be sent to the model.
file_url: optional string or null The URL of the file to be sent to the model.
formaturi filename: optional string or null The name of the file to be sent to the model.
prompt_cache_breakpoint: optional object or null Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" type: "function_call_output" (default: "function_call_output") The type of the function tool call output. Always function_call_output.
id: optional string or null The unique ID of the function tool call output. Populated when this item is returned via API.
call_id: optional string or null The unique ID of the function tool call generated by the model.
minLength1 maxLength64 caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
name: optional string or null The name of the tool that produced the output.
minLength1 maxLength128 namespace: optional string or null The namespace of the tool that produced the output.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 status: optional "in_progress" or "completed" or "incomplete" or null The status of the item. One of in_progress, completed, or incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" ToolSearchCall { arguments, type, id, 3 more } arguments: unknown The arguments supplied to the tool search call.
type: "tool_search_call" (default: "tool_search_call") The item type. Always tool_search_call.
id: optional string or null The unique ID of this tool search call.
call_id: optional string or null The unique ID of the tool search call generated by the model.
minLength1 maxLength64 execution: optional "server" or "client" Whether tool search was executed by the server or by the client.
One of the following: "server" "client" status: optional "in_progress" or "completed" or "incomplete" or null The status of the tool search call.
One of the following: "in_progress" "completed" "incomplete" ToolSearchOutput { tools, type, id, 3 more } tools: array of Function or FileSearch or Computer or 13 more The loaded tool definitions returned by the tool search output.
One of the following: Function { name, parameters, strict, 6 more } Defines a function in your own code the model can choose to call. Learn more about function calling.
name: string The name of the function to call.
parameters: map[unknown] or null A JSON schema object describing the parameters of the function.
strict: boolean or null Whether strict parameter validation is enforced for this function tool.
type: "function" (default: "function") The type of the function tool. Always function.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
One of the following: "direct" "programmatic" async: optional boolean defer_loading: optional boolean Whether this function is deferred and loaded via tool search.
description: optional string or null A description of the function. Used by the model to determine whether or not to call the function.
output_schema: optional map[unknown] or null A JSON schema object describing the JSON value encoded in string outputs for this function.
FileSearch { type, vector_store_ids, filters, 2 more } A tool that searches for relevant content from uploaded files. Learn more about the file search tool.
type: "file_search" (default: "file_search") The type of the file search tool. Always file_search.
vector_store_ids: array of string The IDs of the vector stores to search.
filters: optional ComparisonFilter or CompoundFilter or null A filter to apply.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } Combine multiple filters using and or or.
filters: array of ComparisonFilter or CompoundFilter Array of filters to combine. Items can be ComparisonFilter or CompoundFilter.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } This type repeats the enclosing schema. type: "and" or "or" Type of operation: and or or.
One of the following: "and" "or" max_num_results: optional integer The maximum number of results to return. This number should be between 1 and 50 inclusive.
ranking_options: optional { hybrid_search, ranker, score_threshold } Ranking options for search.
hybrid_search: optional { embedding_weight, text_weight } Weights that control how reciprocal rank fusion balances semantic embedding matches versus sparse keyword matches when hybrid search is enabled.
embedding_weight: number The weight of the embedding in the reciprocal ranking fusion.
text_weight: number The weight of the text in the reciprocal ranking fusion.
ranker: optional "auto" or "default-2024-11-15" The ranker to use for the file search.
One of the following: "auto" "default-2024-11-15" score_threshold: optional number The score threshold for the file search, a number between 0 and 1. Numbers closer to 1 will attempt to return only the most relevant results, but may return fewer results.
Computer { type } A tool that controls a virtual computer. Learn more about the computer tool.
type: "computer" (default: "computer") The type of the computer tool. Always computer.
ComputerUsePreview { display_height, display_width, environment, type } A tool that controls a virtual computer. Learn more about the computer tool.
display_height: integer The height of the computer display.
display_width: integer The width of the computer display.
environment: "windows" or "mac" or "linux" or 2 more The type of computer environment to control.
One of the following: "windows" "mac" "linux" "ubuntu" "browser" type: "computer_use_preview" (default: "computer_use_preview") The type of the computer use tool. Always computer_use_preview.
WebSearch { type, external_web_access, filters, 2 more } Search the Internet for sources related to the prompt. Learn more about the
web search tool.
type: "web_search" or "web_search_2025_08_26" (default: "web_search") The type of the web search tool. One of web_search or web_search_2025_08_26.
One of the following: "web_search" "web_search_2025_08_26" external_web_access: optional boolean (default: true) Allow live internet access for web search. Defaults to true when omitted. When false, the web search tool runs in offline/cache-only mode and will not fetch new external content.
filters: optional object or null Filters for the search.
allowed_domains: optional array of string or null (default: []) Allowed domains for the search. If not provided, all domains are allowed.
Subdomains of the provided domains are allowed as well.
Example: ["pubmed.ncbi.nlm.nih.gov"]
search_context_size: optional "low" or "medium" or "high" (default: "medium") High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the
United States. To avoid this fallback, pass {"type": "approximate"} without
location fields. To localize results, provide the relevant location fields.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
type: optional "approximate" (default: "approximate") The type of location approximation. Always approximate.
Mcp { server_label, type, allowed_callers, 9 more } Give the model access to additional tools via remote Model Context Protocol
(MCP) servers. Learn more about MCP.
server_label: string A label for this MCP server, used to identify it in tool calls.
type: "mcp" The type of the MCP tool. Always mcp.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" allowed_tools: optional array of string or McpToolFilter or null List of allowed tool names or a filter object.
One of the following: McpAllowedTools = array of string A string array of allowed tool names
McpToolFilter { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
authorization: optional string An OAuth access token that can be used with a remote MCP server, either
with a custom MCP server URL or a service connector. Your application
must handle the OAuth authorization flow and provide the token here.
connector_id: optional "connector_dropbox" or "connector_gmail" or "connector_googlecalendar" or 5 more Deprecated.
Identifier for service connectors, like those available in ChatGPT. One of
server_url, connector_id, or tunnel_id must be provided. Learn more
about service connectors here.
This field is deprecated for models released after September 1, 2026.
Use server_url to connect to a remote MCP server, or tunnel_id to
connect through a Secure MCP Tunnel.
Currently supported connector_id values are:
Dropbox: connector_dropbox
Gmail: connector_gmail
Google Calendar: connector_googlecalendar
Google Drive: connector_googledrive
Microsoft Teams: connector_microsoftteams
Outlook Calendar: connector_outlookcalendar
Outlook Email: connector_outlookemail
SharePoint: connector_sharepoint
One of the following: "connector_dropbox" "connector_gmail" "connector_googlecalendar" "connector_googledrive" "connector_microsoftteams" "connector_outlookcalendar" "connector_outlookemail" "connector_sharepoint" defer_loading: optional boolean Whether this MCP tool is deferred and discovered via tool search.
headers: optional map[string] or null Optional HTTP headers to send to the MCP server. Use for authentication
or other purposes.
require_approval: optional McpToolApprovalFilter or "always" or "never" or null (default: "always") Specify which of the MCP server’s tools require approval.
One of the following: McpToolApprovalFilter { always, never } Specify which of the MCP server’s tools require approval. Can be
always, never, or a filter object associated with tools
that require approval.
always: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
never: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
McpToolApprovalSetting = "always" or "never" Specify a single approval policy for all tools. One of always or
never. When set to always, all tools will require approval. When
set to never, all tools will not require approval.
One of the following: "always" "never" server_description: optional string Optional description of the MCP server, used to provide more context.
server_url: optional string The URL for the MCP server. One of server_url, connector_id, or
tunnel_id must be provided.
formaturi tunnel_id: optional string The Secure MCP Tunnel ID to use instead of a direct server URL. One of
server_url, connector_id, or tunnel_id must be provided.
pattern^tunnel_[a-z0-9]{32}$ CodeInterpreter { container, type, allowed_callers } A tool that runs Python code to help generate a response to a prompt.
container: string or CodeInterpreterToolAuto The code interpreter container. Can be a container ID or an object that
specifies uploaded file IDs to make available to your code, along with an
optional memory_limit setting.
One of the following: string The container ID.
CodeInterpreterToolAuto { type, file_ids, memory_limit, network_policy } Configuration for a code interpreter container. Optionally specify the IDs of the files to run the code on.
type: "auto" (default: "auto") Always auto.
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the code interpreter container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 type: "code_interpreter" The type of the code interpreter tool. Always code_interpreter.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" ProgrammaticToolCalling { type } type: "programmatic_tool_calling" (default: "programmatic_tool_calling") The type of the tool. Always programmatic_tool_calling.
ImageGeneration { type, action, background, 13 more } A tool that generates images using the GPT image models.
type: "image_generation" The type of the image generation tool. Always image_generation.
action: optional "generate" or "edit" or "auto" Whether to generate a new image or edit an existing image. Default: auto.
One of the following: "generate" "edit" "auto" background: optional "transparent" or "opaque" or "auto" (default: "auto") Set the background of the generated image. One of transparent, opaque,
or auto. gpt-image-2.5-sunburst and gpt-image-2.5-flare, including
their 2026-09-08 snapshots, support opaque and transparent
backgrounds. Transparent backgrounds are available for supported GPT Image
models. For gpt-image-2 and gpt-image-2-2026-04-21, this support is in
preview. When using transparent, set the output format to png or webp.
Default: auto.
One of the following: "transparent" "opaque" "auto" complexity: optional "low" or "medium" or "high" Computation spent planning the image, independently of rendering quality.
Overrides the outer response’s reasoning.effort for image generation.
When omitted, inherits a supported outer reasoning effort or uses the
image model’s configured default.
For SVG output, controls the reference image, not the SVG author’s
reasoning.
One of the following: "low" "medium" "high" frame_count: optional integer (default: 9) Number of GIF frames. Requires output_format to be gif. Default: 9.
minimum2 maximum16 frame_duration_ms: optional integer (default: 200) Duration of each GIF frame in milliseconds, rounded to 10 ms.
Requires output_format to be gif. Default: 200.
minimum20 maximum2000 input_fidelity: optional "high" or "low" or null Control how much effort the model will exert to match the style and features, especially facial features, of input images. Supports high and low on gpt-image-1.5; gpt-image-1-mini supports only low. For gpt-image-2, omit this parameter. Defaults to low on supported models.
One of the following: "high" "low" input_image_mask: optional { file_id, image_url } Optional mask for inpainting. Contains image_url
(string, optional) and file_id (string, optional).
file_id: optional string File ID for the mask image.
image_url: optional string Base64-encoded mask image.
model: optional string or "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: string "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: "gpt-image-1-mini" "gpt-image-2" "gpt-image-2-2026-04-21" "gpt-image-2.5-sunburst" "gpt-image-2.5-sunburst-2026-09-08" "gpt-image-2.5-flare" "gpt-image-2.5-flare-2026-09-08" "gpt-image-2.6-sunburst" "gpt-image-2.6-flare" "gpt-image-1.5" "chatgpt-image-latest" "gpt-image-1" Deprecated. GPT Image 1 was shut down on October 23, 2026. Use gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
moderation: optional "auto" or "low" (default: "auto") Moderation level for the generated image. Default: auto.
One of the following: "auto" "low" output_compression: optional integer (default: 100) Compression level for the output image. Default: 100.
minimum0 maximum100 output_format: optional "png" or "webp" or "jpeg" or 2 more (default: "png") The output format of the generated image. One of png, webp, jpeg,
gif, or svg. Default: png. GIF and SVG output require a model
that supports the selected format. GIF output does not support
partial image streaming (partial_images greater than 0),
input_image_mask, or custom output_compression.
Responses lifecycle events can still be streamed for GIF output.
One of the following: "png" "webp" "jpeg" "gif" "svg" partial_images: optional integer (default: 0) Number of partial images to generate in streaming mode, from 0 (default value) to 3.
minimum0 maximum3 quality: optional "low" or "medium" or "high" or 3 more (default: "auto") The quality of the generated image. The GPT image models support low,
medium, and high. gpt-image-2.5-sunburst and gpt-image-2.5-flare,
including their 2026-09-08 snapshots, also support xhigh and max.
Default: auto.
One of the following: "low" "medium" "high" "xhigh" "max" "auto" size: optional string or "1024x1024" or "1024x1536" or "1536x1024" or "auto" (default: "auto") The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: string "1024x1024" or "1024x1536" or "1536x1024" or "auto" The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: "1024x1024" "1024x1536" "1536x1024" "auto" svg: optional { model, reasoning } SVG author model settings. Requires output_format to be svg.
When SVG model selection is enabled, omitting this object uses the
outer response model. Otherwise, the default SVG author is used and
explicit model overrides are unavailable.
model: optional string minLength1 reasoning: optional object or null Reasoning effort for SVG authoring. When omitted, uses the author model’s default in both Images and Responses.
effort: optional "none" or "minimal" or "low" or 4 more Controls how much effort the model should spend on reasoning for this request.
One of the following: "none" "minimal" "low" "medium" "high" "xhigh" "max" LocalShell { type } A tool that allows the model to execute shell commands in a local environment.
type: "local_shell" (default: "local_shell") The type of the local shell tool. Always local_shell.
Shell { type, allowed_callers, environment } A tool that allows the model to execute shell commands.
type: "shell" (default: "shell") The type of the shell tool. Always shell.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" environment: optional ContainerAuto or LocalEnvironment or ContainerReference or null One of the following: ContainerAuto { type, file_ids, memory_limit, 2 more } type: "container_auto" (default: "container_auto") Automatically creates a container for this request
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 skills: optional array of SkillReference or InlineSkill An optional list of skills referenced by id or inline data.
maxItems200 One of the following: SkillReference { skill_id, type, version } skill_id: string The ID of the referenced skill.
minLength1 maxLength64 type: "skill_reference" (default: "skill_reference") References a skill created with the /v1/skills endpoint.
version: optional string Optional skill version. Use a positive integer or ‘latest’. Omit for default.
InlineSkill { description, name, source, type } description: string The description of the skill.
name: string The name of the skill.
source: InlineSkillSource { data, media_type, type } Inline skill payload
type: "inline" (default: "inline") Defines an inline skill for this request.
LocalEnvironment { type, skills } type: "local" (default: "local") Use a local computer environment.
skills: optional array of LocalSkill { description, name, path } An optional list of skills.
maxItems200 description: string The description of the skill.
name: string The name of the skill.
path: string The path to the directory containing the skill.
ContainerReference { container_id, type } container_id: string The ID of the referenced container.
type: "container_reference" (default: "container_reference") References a container created with the /v1/containers endpoint
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
ToolSearchOutputNamespaceTool { description, name, tools, type } Groups function/custom tools under a shared namespace.
description: string A description of the namespace shown to the model.
name: string The namespace name used in tool calls (for example, crm).
minLength1 tools: array of Function or Custom The function/custom tools loaded inside this namespace.
minItems1 One of the following: Function { name, type, allowed_callers, 6 more } name: string The name of the loaded function tool.
pattern^(?![\s\S]*[\r\n])[a-zA-Z0-9_.-]+$ minLength1 maxLength128 type: "function" (default: "function") allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this function should be deferred and discovered via tool search.
description: optional string or null output_schema: optional map[unknown] or null A JSON Schema describing the JSON value encoded in string outputs for this function tool. This does not describe content-array outputs.
parameters: optional unknown or null strict: optional boolean or null Whether to enforce strict parameter validation. If omitted, Responses attempts to use strict validation when the schema is compatible, and falls back to non-strict validation otherwise.
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
type: "namespace" (default: "namespace") The type of the tool. Always namespace.
ToolSearch { type, description, execution, parameters } Hosted or BYOT tool search configuration for deferred tools.
type: "tool_search" (default: "tool_search") The type of the tool. Always tool_search.
description: optional string or null Description shown to the model for a client-executed tool search tool.
execution: optional "server" or "client" Whether tool search is executed by the server or by the client.
One of the following: "server" "client" parameters: optional unknown or null Parameter schema for a client-executed tool search tool.
WebSearchPreview { type, search_content_types, search_context_size, user_location } This tool searches the web for relevant results to use in a response. Learn more about the web search tool.
type: "web_search_preview" or "web_search_preview_2025_03_11" (default: "web_search_preview") The type of the web search tool. One of web_search_preview or web_search_preview_2025_03_11.
One of the following: "web_search_preview" "web_search_preview_2025_03_11" search_content_types: optional array of "text" or "image" One of the following: "text" "image" search_context_size: optional "low" or "medium" or "high" High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the United States. To avoid this fallback, pass {"type": "approximate"} without location fields. To localize results, provide the relevant location fields.
type: "approximate" (default: "approximate") The type of location approximation. Always approximate.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
ApplyPatch { type, allowed_callers } Allows the assistant to create, delete, or update files using unified diffs.
type: "apply_patch" (default: "apply_patch") The type of the tool. Always apply_patch.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" type: "tool_search_output" (default: "tool_search_output") The item type. Always tool_search_output.
id: optional string or null The unique ID of this tool search output.
call_id: optional string or null The unique ID of the tool search call generated by the model.
minLength1 maxLength64 execution: optional "server" or "client" Whether tool search was executed by the server or by the client.
One of the following: "server" "client" status: optional "in_progress" or "completed" or "incomplete" or null The status of the tool search output.
One of the following: "in_progress" "completed" "incomplete" AdditionalTools { role, tools, type, id } role: "developer" (default: "developer") The role that provided the additional tools. Only developer is supported.
tools: array of Function or FileSearch or Computer or 13 more A list of additional tools made available at this item.
One of the following: Function { name, parameters, strict, 6 more } Defines a function in your own code the model can choose to call. Learn more about function calling.
name: string The name of the function to call.
parameters: map[unknown] or null A JSON schema object describing the parameters of the function.
strict: boolean or null Whether strict parameter validation is enforced for this function tool.
type: "function" (default: "function") The type of the function tool. Always function.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
One of the following: "direct" "programmatic" async: optional boolean defer_loading: optional boolean Whether this function is deferred and loaded via tool search.
description: optional string or null A description of the function. Used by the model to determine whether or not to call the function.
output_schema: optional map[unknown] or null A JSON schema object describing the JSON value encoded in string outputs for this function.
FileSearch { type, vector_store_ids, filters, 2 more } A tool that searches for relevant content from uploaded files. Learn more about the file search tool.
type: "file_search" (default: "file_search") The type of the file search tool. Always file_search.
vector_store_ids: array of string The IDs of the vector stores to search.
filters: optional ComparisonFilter or CompoundFilter or null A filter to apply.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } Combine multiple filters using and or or.
filters: array of ComparisonFilter or CompoundFilter Array of filters to combine. Items can be ComparisonFilter or CompoundFilter.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } This type repeats the enclosing schema. type: "and" or "or" Type of operation: and or or.
One of the following: "and" "or" max_num_results: optional integer The maximum number of results to return. This number should be between 1 and 50 inclusive.
ranking_options: optional { hybrid_search, ranker, score_threshold } Ranking options for search.
hybrid_search: optional { embedding_weight, text_weight } Weights that control how reciprocal rank fusion balances semantic embedding matches versus sparse keyword matches when hybrid search is enabled.
embedding_weight: number The weight of the embedding in the reciprocal ranking fusion.
text_weight: number The weight of the text in the reciprocal ranking fusion.
ranker: optional "auto" or "default-2024-11-15" The ranker to use for the file search.
One of the following: "auto" "default-2024-11-15" score_threshold: optional number The score threshold for the file search, a number between 0 and 1. Numbers closer to 1 will attempt to return only the most relevant results, but may return fewer results.
Computer { type } A tool that controls a virtual computer. Learn more about the computer tool.
type: "computer" (default: "computer") The type of the computer tool. Always computer.
ComputerUsePreview { display_height, display_width, environment, type } A tool that controls a virtual computer. Learn more about the computer tool.
display_height: integer The height of the computer display.
display_width: integer The width of the computer display.
environment: "windows" or "mac" or "linux" or 2 more The type of computer environment to control.
One of the following: "windows" "mac" "linux" "ubuntu" "browser" type: "computer_use_preview" (default: "computer_use_preview") The type of the computer use tool. Always computer_use_preview.
WebSearch { type, external_web_access, filters, 2 more } Search the Internet for sources related to the prompt. Learn more about the
web search tool.
type: "web_search" or "web_search_2025_08_26" (default: "web_search") The type of the web search tool. One of web_search or web_search_2025_08_26.
One of the following: "web_search" "web_search_2025_08_26" external_web_access: optional boolean (default: true) Allow live internet access for web search. Defaults to true when omitted. When false, the web search tool runs in offline/cache-only mode and will not fetch new external content.
filters: optional object or null Filters for the search.
allowed_domains: optional array of string or null (default: []) Allowed domains for the search. If not provided, all domains are allowed.
Subdomains of the provided domains are allowed as well.
Example: ["pubmed.ncbi.nlm.nih.gov"]
search_context_size: optional "low" or "medium" or "high" (default: "medium") High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the
United States. To avoid this fallback, pass {"type": "approximate"} without
location fields. To localize results, provide the relevant location fields.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
type: optional "approximate" (default: "approximate") The type of location approximation. Always approximate.
Mcp { server_label, type, allowed_callers, 9 more } Give the model access to additional tools via remote Model Context Protocol
(MCP) servers. Learn more about MCP.
server_label: string A label for this MCP server, used to identify it in tool calls.
type: "mcp" The type of the MCP tool. Always mcp.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" allowed_tools: optional array of string or McpToolFilter or null List of allowed tool names or a filter object.
One of the following: McpAllowedTools = array of string A string array of allowed tool names
McpToolFilter { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
authorization: optional string An OAuth access token that can be used with a remote MCP server, either
with a custom MCP server URL or a service connector. Your application
must handle the OAuth authorization flow and provide the token here.
connector_id: optional "connector_dropbox" or "connector_gmail" or "connector_googlecalendar" or 5 more Deprecated.
Identifier for service connectors, like those available in ChatGPT. One of
server_url, connector_id, or tunnel_id must be provided. Learn more
about service connectors here.
This field is deprecated for models released after September 1, 2026.
Use server_url to connect to a remote MCP server, or tunnel_id to
connect through a Secure MCP Tunnel.
Currently supported connector_id values are:
Dropbox: connector_dropbox
Gmail: connector_gmail
Google Calendar: connector_googlecalendar
Google Drive: connector_googledrive
Microsoft Teams: connector_microsoftteams
Outlook Calendar: connector_outlookcalendar
Outlook Email: connector_outlookemail
SharePoint: connector_sharepoint
One of the following: "connector_dropbox" "connector_gmail" "connector_googlecalendar" "connector_googledrive" "connector_microsoftteams" "connector_outlookcalendar" "connector_outlookemail" "connector_sharepoint" defer_loading: optional boolean Whether this MCP tool is deferred and discovered via tool search.
headers: optional map[string] or null Optional HTTP headers to send to the MCP server. Use for authentication
or other purposes.
require_approval: optional McpToolApprovalFilter or "always" or "never" or null (default: "always") Specify which of the MCP server’s tools require approval.
One of the following: McpToolApprovalFilter { always, never } Specify which of the MCP server’s tools require approval. Can be
always, never, or a filter object associated with tools
that require approval.
always: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
never: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
McpToolApprovalSetting = "always" or "never" Specify a single approval policy for all tools. One of always or
never. When set to always, all tools will require approval. When
set to never, all tools will not require approval.
One of the following: "always" "never" server_description: optional string Optional description of the MCP server, used to provide more context.
server_url: optional string The URL for the MCP server. One of server_url, connector_id, or
tunnel_id must be provided.
formaturi tunnel_id: optional string The Secure MCP Tunnel ID to use instead of a direct server URL. One of
server_url, connector_id, or tunnel_id must be provided.
pattern^tunnel_[a-z0-9]{32}$ CodeInterpreter { container, type, allowed_callers } A tool that runs Python code to help generate a response to a prompt.
container: string or CodeInterpreterToolAuto The code interpreter container. Can be a container ID or an object that
specifies uploaded file IDs to make available to your code, along with an
optional memory_limit setting.
One of the following: string The container ID.
CodeInterpreterToolAuto { type, file_ids, memory_limit, network_policy } Configuration for a code interpreter container. Optionally specify the IDs of the files to run the code on.
type: "auto" (default: "auto") Always auto.
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the code interpreter container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 type: "code_interpreter" The type of the code interpreter tool. Always code_interpreter.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" ProgrammaticToolCalling { type } type: "programmatic_tool_calling" (default: "programmatic_tool_calling") The type of the tool. Always programmatic_tool_calling.
ImageGeneration { type, action, background, 13 more } A tool that generates images using the GPT image models.
type: "image_generation" The type of the image generation tool. Always image_generation.
action: optional "generate" or "edit" or "auto" Whether to generate a new image or edit an existing image. Default: auto.
One of the following: "generate" "edit" "auto" background: optional "transparent" or "opaque" or "auto" (default: "auto") Set the background of the generated image. One of transparent, opaque,
or auto. gpt-image-2.5-sunburst and gpt-image-2.5-flare, including
their 2026-09-08 snapshots, support opaque and transparent
backgrounds. Transparent backgrounds are available for supported GPT Image
models. For gpt-image-2 and gpt-image-2-2026-04-21, this support is in
preview. When using transparent, set the output format to png or webp.
Default: auto.
One of the following: "transparent" "opaque" "auto" complexity: optional "low" or "medium" or "high" Computation spent planning the image, independently of rendering quality.
Overrides the outer response’s reasoning.effort for image generation.
When omitted, inherits a supported outer reasoning effort or uses the
image model’s configured default.
For SVG output, controls the reference image, not the SVG author’s
reasoning.
One of the following: "low" "medium" "high" frame_count: optional integer (default: 9) Number of GIF frames. Requires output_format to be gif. Default: 9.
minimum2 maximum16 frame_duration_ms: optional integer (default: 200) Duration of each GIF frame in milliseconds, rounded to 10 ms.
Requires output_format to be gif. Default: 200.
minimum20 maximum2000 input_fidelity: optional "high" or "low" or null Control how much effort the model will exert to match the style and features, especially facial features, of input images. Supports high and low on gpt-image-1.5; gpt-image-1-mini supports only low. For gpt-image-2, omit this parameter. Defaults to low on supported models.
One of the following: "high" "low" input_image_mask: optional { file_id, image_url } Optional mask for inpainting. Contains image_url
(string, optional) and file_id (string, optional).
file_id: optional string File ID for the mask image.
image_url: optional string Base64-encoded mask image.
model: optional string or "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: string "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: "gpt-image-1-mini" "gpt-image-2" "gpt-image-2-2026-04-21" "gpt-image-2.5-sunburst" "gpt-image-2.5-sunburst-2026-09-08" "gpt-image-2.5-flare" "gpt-image-2.5-flare-2026-09-08" "gpt-image-2.6-sunburst" "gpt-image-2.6-flare" "gpt-image-1.5" "chatgpt-image-latest" "gpt-image-1" Deprecated. GPT Image 1 was shut down on October 23, 2026. Use gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
moderation: optional "auto" or "low" (default: "auto") Moderation level for the generated image. Default: auto.
One of the following: "auto" "low" output_compression: optional integer (default: 100) Compression level for the output image. Default: 100.
minimum0 maximum100 output_format: optional "png" or "webp" or "jpeg" or 2 more (default: "png") The output format of the generated image. One of png, webp, jpeg,
gif, or svg. Default: png. GIF and SVG output require a model
that supports the selected format. GIF output does not support
partial image streaming (partial_images greater than 0),
input_image_mask, or custom output_compression.
Responses lifecycle events can still be streamed for GIF output.
One of the following: "png" "webp" "jpeg" "gif" "svg" partial_images: optional integer (default: 0) Number of partial images to generate in streaming mode, from 0 (default value) to 3.
minimum0 maximum3 quality: optional "low" or "medium" or "high" or 3 more (default: "auto") The quality of the generated image. The GPT image models support low,
medium, and high. gpt-image-2.5-sunburst and gpt-image-2.5-flare,
including their 2026-09-08 snapshots, also support xhigh and max.
Default: auto.
One of the following: "low" "medium" "high" "xhigh" "max" "auto" size: optional string or "1024x1024" or "1024x1536" or "1536x1024" or "auto" (default: "auto") The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: string "1024x1024" or "1024x1536" or "1536x1024" or "auto" The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: "1024x1024" "1024x1536" "1536x1024" "auto" svg: optional { model, reasoning } SVG author model settings. Requires output_format to be svg.
When SVG model selection is enabled, omitting this object uses the
outer response model. Otherwise, the default SVG author is used and
explicit model overrides are unavailable.
model: optional string minLength1 reasoning: optional object or null Reasoning effort for SVG authoring. When omitted, uses the author model’s default in both Images and Responses.
effort: optional "none" or "minimal" or "low" or 4 more Controls how much effort the model should spend on reasoning for this request.
One of the following: "none" "minimal" "low" "medium" "high" "xhigh" "max" LocalShell { type } A tool that allows the model to execute shell commands in a local environment.
type: "local_shell" (default: "local_shell") The type of the local shell tool. Always local_shell.
Shell { type, allowed_callers, environment } A tool that allows the model to execute shell commands.
type: "shell" (default: "shell") The type of the shell tool. Always shell.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" environment: optional ContainerAuto or LocalEnvironment or ContainerReference or null One of the following: ContainerAuto { type, file_ids, memory_limit, 2 more } type: "container_auto" (default: "container_auto") Automatically creates a container for this request
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 skills: optional array of SkillReference or InlineSkill An optional list of skills referenced by id or inline data.
maxItems200 One of the following: SkillReference { skill_id, type, version } skill_id: string The ID of the referenced skill.
minLength1 maxLength64 type: "skill_reference" (default: "skill_reference") References a skill created with the /v1/skills endpoint.
version: optional string Optional skill version. Use a positive integer or ‘latest’. Omit for default.
InlineSkill { description, name, source, type } description: string The description of the skill.
name: string The name of the skill.
source: InlineSkillSource { data, media_type, type } Inline skill payload
type: "inline" (default: "inline") Defines an inline skill for this request.
LocalEnvironment { type, skills } type: "local" (default: "local") Use a local computer environment.
skills: optional array of LocalSkill { description, name, path } An optional list of skills.
maxItems200 description: string The description of the skill.
name: string The name of the skill.
path: string The path to the directory containing the skill.
ContainerReference { container_id, type } container_id: string The ID of the referenced container.
type: "container_reference" (default: "container_reference") References a container created with the /v1/containers endpoint
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
Namespace { description, name, tools, type } Groups function/custom tools under a shared namespace.
description: string A description of the namespace shown to the model.
name: string The namespace name used in tool calls (for example, crm).
minLength1 tools: array of Function or Custom The function/custom tools available inside this namespace.
minItems1 One of the following: Function { name, type, allowed_callers, 6 more } name: string pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 type: "function" (default: "function") allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this function should be deferred and discovered via tool search.
description: optional string or null output_schema: optional map[unknown] or null A JSON Schema describing the JSON value encoded in string outputs for this function tool. This does not describe content-array outputs.
parameters: optional unknown or null strict: optional boolean or null Whether to enforce strict parameter validation. If omitted, Responses attempts to use strict validation when the schema is compatible, and falls back to non-strict validation otherwise.
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
type: "namespace" (default: "namespace") The type of the tool. Always namespace.
ToolSearch { type, description, execution, parameters } Hosted or BYOT tool search configuration for deferred tools.
type: "tool_search" (default: "tool_search") The type of the tool. Always tool_search.
description: optional string or null Description shown to the model for a client-executed tool search tool.
execution: optional "server" or "client" Whether tool search is executed by the server or by the client.
One of the following: "server" "client" parameters: optional unknown or null Parameter schema for a client-executed tool search tool.
WebSearchPreview { type, search_content_types, search_context_size, user_location } This tool searches the web for relevant results to use in a response. Learn more about the web search tool.
type: "web_search_preview" or "web_search_preview_2025_03_11" (default: "web_search_preview") The type of the web search tool. One of web_search_preview or web_search_preview_2025_03_11.
One of the following: "web_search_preview" "web_search_preview_2025_03_11" search_content_types: optional array of "text" or "image" One of the following: "text" "image" search_context_size: optional "low" or "medium" or "high" High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the United States. To avoid this fallback, pass {"type": "approximate"} without location fields. To localize results, provide the relevant location fields.
type: "approximate" (default: "approximate") The type of location approximation. Always approximate.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
ApplyPatch { type, allowed_callers } Allows the assistant to create, delete, or update files using unified diffs.
type: "apply_patch" (default: "apply_patch") The type of the tool. Always apply_patch.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" type: "additional_tools" (default: "additional_tools") The item type. Always additional_tools.
id: optional string or null The unique ID of this additional tools item.
ConfigurationUpdate { type, id, reasoning, text } An update to the conversation’s response configuration. Supply reasoning,
text, or both. Each setting stays in effect until a later update replaces it.
type: "configuration_update" (default: "configuration_update") The item type. Always configuration_update.
id: optional string or null The unique ID of the configuration update item.
reasoning: optional { effort } Updates to reasoning configuration. Only effort is supported.
effort: optional ReasoningEffort or null (default: "medium") The reasoning effort to use for subsequent responses until another
configuration update replaces it.
text: optional { format } The text output configuration for subsequent responses.
format: ResponseFormatText or ResponseFormatTextJSONSchemaConfig Use a JSON schema for structured output or text to return to plain text.
One of the following: ResponseFormatText { type } Default response format. Used to generate text responses.
type: "text" The type of response format being defined. Always text.
ResponseFormatTextJSONSchemaConfig { name, schema, type, 2 more } JSON Schema response format. Used to generate structured JSON responses.
Learn more about Structured Outputs.
name: string The name of the response format. Must be a-z, A-Z, 0-9, or contain
underscores and dashes, with a maximum length of 64.
schema: map[unknown] The schema for the response format, described as a JSON Schema object.
Learn how to build JSON schemas here.
type: "json_schema" The type of response format being defined. Always json_schema.
description: optional string A description of what the response format is for, used by the model to
determine how to respond in the format.
strict: optional boolean or null (default: false) Whether to enable strict schema adherence when generating the output.
If set to true, the model will always follow the exact schema defined
in the schema field. Only a subset of JSON Schema is supported when
strict is true. To learn more, read the Structured Outputs
guide.
Reasoning { id, summary, type, 3 more } A description of the chain of thought used by a reasoning model while generating
a response. Be sure to include these items in your input to the Responses API
for subsequent turns of a conversation if you are manually
managing context.
id: string The unique identifier of the reasoning content.
summary: array of SummaryTextContent { text, type } Reasoning summary content.
text: string A summary of the reasoning output from the model so far.
type: "summary_text" (default: "summary_text") The type of the object. Always summary_text.
type: "reasoning" The type of the object. Always reasoning.
content: optional array of { text, type } Reasoning text content.
text: string The reasoning text from the model.
type: "reasoning_text" (default: "reasoning_text") The type of the reasoning text. Always reasoning_text.
encrypted_content: optional string or null The encrypted content of the reasoning item. This is populated by default
for reasoning items returned by POST /v1/responses and WebSocket
response.create requests.
When streaming, use the completed reasoning item and its
encrypted_content from the response.output_item.done event in
subsequent requests. The encrypted_content in
response.output_item.added may be incomplete. This is especially
important when store is false or when using Zero Data Retention.
status: optional "in_progress" or "completed" or "incomplete" The status of the item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" Compaction { encrypted_content, type, id } A compaction item generated by the v1/responses/compact API.
encrypted_content: string The encrypted content of the compaction summary.
maxLength104857600 type: "compaction" (default: "compaction") The type of the item. Always compaction.
id: optional string or null The ID of the compaction item.
ImageGenerationCall { id, result, status, 9 more } An image generation request made by the model.
id: string The unique ID of the image generation call.
result: string or null The generated image encoded in base64.
status: "in_progress" or "completed" or "generating" or "failed" The status of the image generation call.
One of the following: "in_progress" "completed" "generating" "failed" type: "image_generation_call" The type of the image generation call. Always image_generation_call.
action: optional "generate" or "edit" or "auto" or null The action used for image generation.
One of the following: "generate" "edit" "auto" background: optional "transparent" or "opaque" or "auto" or null The background setting used for generation.
One of the following: "transparent" "opaque" "auto" error: optional { code, message, param, 2 more } Details of a moderation failure, when available.
code: string or null The error code.
message: string A description of the error.
param: string or null The related request parameter.
type: string The error type.
moderation_details: optional { categories, moderation_stage } Public details for a moderation_blocked error, when available.
categories: array of string Public moderation categories. New category values may be added over time.
moderation_stage: "input" or "output" or "unknown" The stage where moderation blocked the image generation call.
One of the following: "input" "output" "unknown" output_format: optional "png" or "webp" or "jpeg" or 2 more or null The output format used for generation.
One of the following: "png" "webp" "jpeg" "gif" "svg" quality: optional "low" or "medium" or "high" or 3 more or null The quality of the image generated by the image generation tool call. One of low, medium, high, xhigh, max, or auto.
One of the following: "low" "medium" "high" "xhigh" "max" "auto" revised_prompt: optional string or null The prompt that was used after any model prompt rewriting.
size: optional string or "1024x1024" or "1024x1536" or "1536x1024" or null The image dimensions as a WIDTHxHEIGHT string, for example 1536x864.
One of the following: string "1024x1024" or "1024x1536" or "1536x1024" The image dimensions as a WIDTHxHEIGHT string, for example 1536x864.
One of the following: "1024x1024" "1024x1536" "1536x1024" svg_usage: optional { input_tokens, model, output_tokens, 2 more } input_tokens: integer model: string output_tokens: integer reasoning_tokens: integer total_tokens: integer CodeInterpreterCall { id, code, container_id, 3 more } A tool call to run code.
id: string The unique ID of the code interpreter tool call.
code: string or null The code to run, or null if not available.
container_id: string The ID of the container used to run the code.
outputs: array of Logs or Image or null The outputs generated by the code interpreter, such as logs or images.
Can be null if no outputs are available.
One of the following: Logs { logs, type } The logs output from the code interpreter.
logs: string The logs output from the code interpreter.
type: "logs" (default: "logs") The type of the output. Always logs.
Image { type, url } The image output from the code interpreter.
type: "image" (default: "image") The type of the output. Always image.
url: string The URL of the image output from the code interpreter.
formaturi status: "in_progress" or "completed" or "incomplete" or 2 more The status of the code interpreter tool call. Valid values are in_progress, completed, incomplete, interpreting, and failed.
One of the following: "in_progress" "completed" "incomplete" "interpreting" "failed" type: "code_interpreter_call" (default: "code_interpreter_call") The type of the code interpreter tool call. Always code_interpreter_call.
LocalShellCall { id, action, call_id, 2 more } A tool call to run a command on the local shell.
id: string The unique ID of the local shell call.
action: { command, env, type, 3 more } Execute a shell command on the server.
command: array of string The command to run.
env: map[string] Environment variables to set for the command.
type: "exec" (default: "exec") The type of the local shell action. Always exec.
timeout_ms: optional integer or null Optional timeout in milliseconds for the command.
user: optional string or null Optional user to run the command as.
working_directory: optional string or null Optional working directory to run the command in.
call_id: string The unique ID of the local shell tool call generated by the model.
status: "in_progress" or "completed" or "incomplete" The status of the local shell call.
One of the following: "in_progress" "completed" "incomplete" type: "local_shell_call" The type of the local shell call. Always local_shell_call.
LocalShellCallOutput { id, output, type, status } The output of a local shell tool call.
id: string The unique ID of the local shell tool call generated by the model.
output: string A JSON string of the output of the local shell tool call.
type: "local_shell_call_output" The type of the local shell tool call output. Always local_shell_call_output.
status: optional "in_progress" or "completed" or "incomplete" or null The status of the item. One of in_progress, completed, or incomplete.
One of the following: "in_progress" "completed" "incomplete" ShellCall { action, call_id, type, 4 more } A tool representing a request to execute one or more shell commands.
action: { commands, max_output_length, timeout_ms } The shell commands and limits that describe how to run the tool call.
commands: array of string Ordered shell commands for the execution environment to run.
max_output_length: optional integer or null Maximum number of UTF-8 characters to capture from combined stdout and stderr output.
timeout_ms: optional integer or null Maximum wall-clock time in milliseconds to allow the shell commands to run.
call_id: string The unique ID of the shell tool call generated by the model.
minLength1 maxLength64 type: "shell_call" (default: "shell_call") The type of the item. Always shell_call.
id: optional string or null The unique ID of the shell tool call. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
environment: optional LocalEnvironment or ContainerReference or null The environment to execute the shell commands in.
One of the following: LocalEnvironment { type, skills } type: "local" (default: "local") Use a local computer environment.
skills: optional array of LocalSkill { description, name, path } An optional list of skills.
maxItems200 description: string The description of the skill.
name: string The name of the skill.
path: string The path to the directory containing the skill.
ContainerReference { container_id, type } container_id: string The ID of the referenced container.
type: "container_reference" (default: "container_reference") References a container created with the /v1/containers endpoint
status: optional "in_progress" or "completed" or "incomplete" or null The status of the shell call. One of in_progress, completed, or incomplete.
One of the following: "in_progress" "completed" "incomplete" ShellCallOutput { call_id, output, type, 4 more } The streamed output items emitted by a shell tool call.
call_id: string The unique ID of the shell tool call generated by the model.
minLength1 maxLength64 output: array of ResponseFunctionShellCallOutputContent { outcome, stderr, stdout } Captured chunks of stdout and stderr output, along with their associated outcomes.
outcome: Timeout or Exit The exit or timeout outcome associated with this shell call.
One of the following: Timeout { type } Indicates that the shell call exceeded its configured time limit.
type: "timeout" (default: "timeout") The outcome type. Always timeout.
Exit { exit_code, type } Indicates that the shell commands finished and returned an exit code.
exit_code: integer The exit code returned by the shell process.
type: "exit" (default: "exit") The outcome type. Always exit.
stderr: string Captured stderr output for the shell call.
maxLength10485760 stdout: string Captured stdout output for the shell call.
maxLength10485760 type: "shell_call_output" (default: "shell_call_output") The type of the item. Always shell_call_output.
id: optional string or null The unique ID of the shell tool call output. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
max_output_length: optional integer or null The maximum number of UTF-8 characters captured for this shell call’s combined output.
status: optional "in_progress" or "completed" or "incomplete" or null The status of the shell call output.
One of the following: "in_progress" "completed" "incomplete" ApplyPatchCall { call_id, operation, status, 3 more } A tool call representing a request to create, delete, or update files using diff patches.
call_id: string The unique ID of the apply patch tool call generated by the model.
minLength1 maxLength64 operation: CreateFile or DeleteFile or UpdateFile The specific create, delete, or update instruction for the apply_patch tool call.
One of the following: CreateFile { diff, path, type } Instruction for creating a new file via the apply_patch tool.
diff: string Unified diff content to apply when creating the file.
maxLength10485760 path: string Path of the file to create relative to the workspace root.
minLength1 type: "create_file" (default: "create_file") The operation type. Always create_file.
DeleteFile { path, type } Instruction for deleting an existing file via the apply_patch tool.
path: string Path of the file to delete relative to the workspace root.
minLength1 type: "delete_file" (default: "delete_file") The operation type. Always delete_file.
UpdateFile { diff, path, type } Instruction for updating an existing file via the apply_patch tool.
diff: string Unified diff content to apply to the existing file.
maxLength10485760 path: string Path of the file to update relative to the workspace root.
minLength1 type: "update_file" (default: "update_file") The operation type. Always update_file.
status: "in_progress" or "completed" The status of the apply patch tool call. One of in_progress or completed.
One of the following: "in_progress" "completed" type: "apply_patch_call" (default: "apply_patch_call") The type of the item. Always apply_patch_call.
id: optional string or null The unique ID of the apply patch tool call. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
ApplyPatchCallOutput { call_id, status, type, 3 more } The streamed output emitted by an apply patch tool call.
call_id: string The unique ID of the apply patch tool call generated by the model.
minLength1 maxLength64 status: "completed" or "failed" The status of the apply patch tool call output. One of completed or failed.
One of the following: "completed" "failed" type: "apply_patch_call_output" (default: "apply_patch_call_output") The type of the item. Always apply_patch_call_output.
id: optional string or null The unique ID of the apply patch tool call output. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
output: optional string or null Optional human-readable log text from the apply patch tool (e.g., patch results or errors).
maxLength10485760 McpListTools { id, server_label, tools, 2 more } A list of tools available on an MCP server.
id: string The unique ID of the list.
server_label: string The label of the MCP server.
tools: array of { input_schema, name, annotations, description } The tools available on the server.
input_schema: unknown The JSON schema describing the tool’s input.
name: string The name of the tool.
annotations: optional unknown or null Additional annotations about the tool.
description: optional string or null The description of the tool.
type: "mcp_list_tools" The type of the item. Always mcp_list_tools.
error: optional string or null Error message if the server could not list tools.
McpApprovalRequest { id, arguments, name, 2 more } A request for human approval of a tool invocation.
id: string The unique ID of the approval request.
arguments: string A JSON string of arguments for the tool.
name: string The name of the tool to run.
server_label: string The label of the MCP server making the request.
type: "mcp_approval_request" The type of the item. Always mcp_approval_request.
McpApprovalResponse { approval_request_id, approve, type, 2 more } A response to an MCP approval request.
approval_request_id: string The ID of the approval request being answered.
approve: boolean Whether the request was approved.
type: "mcp_approval_response" The type of the item. Always mcp_approval_response.
id: optional string or null The unique ID of the approval response
reason: optional string or null Optional reason for the decision.
McpCall { id, arguments, name, 6 more } An invocation of a tool on an MCP server.
id: string The unique ID of the tool call.
arguments: string A JSON string of the arguments passed to the tool.
name: string The name of the tool that was run.
server_label: string The label of the MCP server running the tool.
type: "mcp_call" The type of the item. Always mcp_call.
approval_request_id: optional string or null Unique identifier for the MCP tool call approval request.
Include this value in a subsequent mcp_approval_response input to approve or reject the corresponding tool call.
error: optional McpToolCallError or null The error from the tool call, if any.
output: optional string or null The output from the tool call.
status: optional "in_progress" or "completed" or "incomplete" or 2 more The status of the tool call. One of in_progress, completed, incomplete, calling, or failed.
One of the following: "in_progress" "completed" "incomplete" "calling" "failed" CustomToolCallOutput { call_id, output, type, 2 more } The output of a custom tool call from your code, being sent back to the model.
call_id: string The call ID, used to map this custom tool call output to a custom tool call.
output: string or array of ResponseInputText or ResponseInputImage or ResponseInputFile The output from the custom tool call generated by your code.
Can be a string or an list of output content.
One of the following: StringOutput = string A string of the output of the custom tool call.
OutputContentList = array of ResponseInputText or ResponseInputImage or ResponseInputFile Text, image, or file output of the custom tool call.
One of the following: ResponseInputText { text, type, prompt_cache_breakpoint } A text input to the model.
text: string The text input to the model.
type: "input_text" (default: "input_text") The type of the input item. Always input_text.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputImage { detail, type, file_id, 2 more } An image input to the model. Learn about image inputs.
detail: ImageDetail The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
type: "input_image" (default: "input_image") The type of the input item. Always input_image.
file_id: optional string or null The ID of the file to be sent to the model.
image_url: optional string or null The URL of the image to be sent to the model. A fully qualified URL or base64 encoded image in a data URL.
formaturi prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputFile { type, detail, file_data, 4 more } A file input to the model.
type: "input_file" (default: "input_file") The type of the input item. Always input_file.
detail: optional "auto" or "low" or "high" The detail level of the file to be sent to the model. Use auto to let the system select the detail level; for GPT-5.6 and later models, auto uses high-quality rendering, which may increase input token usage. Use low for lower-cost rendering, or high to render the file at higher quality. Defaults to auto.
One of the following: "auto" "low" "high" file_data: optional string The content of the file to be sent to the model.
file_id: optional string or null The ID of the file to be sent to the model.
file_url: optional string The URL of the file to be sent to the model.
formaturi filename: optional string The name of the file to be sent to the model.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" type: "custom_tool_call_output" The type of the custom tool call output. Always custom_tool_call_output.
id: optional string The unique ID of the custom tool call output in the OpenAI platform.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
CustomToolCall { call_id, input, name, 5 more } A call to a custom tool created by the model.
call_id: string An identifier used to map this custom tool call to a tool call output.
input: string The input for the custom tool call generated by the model.
name: string The name of the custom tool being called.
type: "custom_tool_call" The type of the custom tool call. Always custom_tool_call.
id: optional string The unique ID of the custom tool call in the OpenAI platform.
async: optional boolean Whether the custom tool call runs asynchronously.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
type: "program" (default: "program") namespace: optional string The namespace of the custom tool being called.
CompactionTrigger { type, id } Compacts the current context. Must be the final input item.
type: "compaction_trigger" (default: "compaction_trigger") The type of the item. Always compaction_trigger.
id: optional string or null The unique ID of this compaction trigger.
ItemReference { id, type } An internal identifier for an item to reference.
id: string The ID of the item to reference.
type: optional "item_reference" or null (default: "item_reference") The type of item to reference. Always item_reference.
Program { id, call_id, code, 2 more } id: string The unique ID of this program item.
call_id: string The stable call ID of the program item.
minLength1 maxLength64 code: string The JavaScript source executed by programmatic tool calling.
maxLength10485760 fingerprint: string Opaque program replay fingerprint that must be round-tripped.
maxLength10485760 type: "program" (default: "program") The item type. Always program.
ProgramOutput { id, call_id, result, 2 more } id: string The unique ID of this program output item.
call_id: string The call ID of the program item.
minLength1 maxLength64 result: string The result produced by the program item.
maxLength10485760 status: "completed" or "incomplete" The terminal status of the program output.
One of the following: "completed" "incomplete" type: "program_output" (default: "program_output") The item type. Always program_output.
type: "response.item.create" (default: "response.item.create") The Live client event type. Always response.item.create.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 ResponseCreateEvent { type, event_id } Request a response from the Live session’s Responses backend, or continue a delegated response waiting for tool results. Requires Responses delegation.
type: "response.create" (default: "response.create") The Live client event type. Always response.create.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 SessionCloseEvent { type, event_id } Request that the Live session close. The terminal session.closed event contains the close reason and final usage.
type: "session.close" (default: "session.close") The Live client event type. Always session.close.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512
ForkServerEvent = SessionStartedEvent { event_id, session, type, client_event_id } or SessionUpdatedEvent { event_id, session, type, client_event_id } or InputAudioMutedEvent { event_id, type, client_event_id } or 24 moreServer events for Live. Response lifecycle events are wrapped inside response.event; dispatch the nested event by its full type and tolerate new response event types. Follow the Live prompting guide when designing the conversation and delegation policy.
ForkServerEvent = SessionStartedEvent { event_id, session, type, client_event_id } or SessionUpdatedEvent { event_id, session, type, client_event_id } or InputAudioMutedEvent { event_id, type, client_event_id } or 24 more
Server events for Live. Response lifecycle events are wrapped inside response.event; dispatch the nested event by its full type and tolerate new response event types. Follow the Live prompting guide when designing the conversation and delegation policy.
One of the following: SessionStartedEvent { event_id, session, type, client_event_id } Returned when a Live session has started. Contains the resolved session configuration, including server defaults.
event_id: string The unique ID of the Live server event.
session: SessionResource { id, expires_at, model, 9 more } The resolved Live session configuration and server-assigned session metadata.
type: "session.started" (default: "session.started") The event type, always session.started.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
SessionUpdatedEvent { event_id, session, type, client_event_id } Returned when a Live session update is accepted. Contains the resolved session configuration after the update.
event_id: string The unique ID of the Live server event.
session: SessionResource { id, expires_at, model, 9 more } The resolved Live session configuration and server-assigned session metadata.
type: "session.updated" (default: "session.updated") The event type, always session.updated.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InputAudioMutedEvent { event_id, type, client_event_id } Acknowledges a session.input_audio.mute command. An independent input mute sends silence to the model and preserves sideband audio reflection. Fixture playback does not emit this event.
event_id: string The unique ID of the Live server event.
type: "session.input_audio.muted" (default: "session.input_audio.muted") The event type, always session.input_audio.muted.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InputAudioUnmutedEvent { event_id, type, client_event_id } Acknowledges a session.input_audio.unmute command, removing the independent input mute. An active audio fixture still pauses inference until playback finishes. Fixture playback does not emit this event.
event_id: string The unique ID of the Live server event.
type: "session.input_audio.unmuted" (default: "session.input_audio.unmuted") The event type, always session.input_audio.unmuted.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
OutputAudioFixturePlayingEvent { event_id, type, client_event_id } The existing playback path has accepted the validated audio fixture. This acknowledgment does not identify when the listener receives the first sample. The optional client_event_id echoes the play command.
event_id: string The unique ID of the Live server event.
type: "session.output_audio.fixture.playing" (default: "session.output_audio.fixture.playing") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
OutputAudioFixturePlayedEvent { event_id, type, client_event_id } The complete audio fixture has drained from the server mixer and normal inference has resumed. Transport buffers may still contain audio; this event does not prove listener-side completion. The optional client_event_id echoes the play command. This event is not emitted for canceled or failed playback.
event_id: string The unique ID of the Live server event.
type: "session.output_audio.fixture.played" (default: "session.output_audio.fixture.played") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InstructionsAppendedEvent { end_ms, event_id, start_ms, 2 more } Returned when a session.instructions.append command is accepted into the Live session timeline. Acknowledges the appended instructions without guaranteeing that the model has acted on them.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.instructions.appended" (default: "session.instructions.appended") The event type, always session.instructions.appended.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
ThinkingAppendedEvent { end_ms, event_id, start_ms, 2 more } Returned when a session.thinking.append command is accepted into the Live session timeline. Acknowledges the added reasoning context without guaranteeing any spoken output.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.thinking.appended" (default: "session.thinking.appended") The event type, always session.thinking.appended.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
CommentaryAppendedEvent { end_ms, event_id, start_ms, 2 more } Returned when a session.commentary.append command is accepted into the Live session timeline. Acknowledges the added commentary without guaranteeing exact wording or completed audio playback.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.commentary.appended" (default: "session.commentary.appended") The event type, always session.commentary.appended.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
SessionInputAudioAppend { audio, type } Input audio received from the primary transport and reflected to a Live sideband connection before model-input muting.
audio: string Base64-encoded raw mono PCM16LE at 24 kHz received from the primary transport, reflected to the sideband before model-input muting. This server event uses the same audio key as the client command, but is not an acknowledgment of it.
type: "session.input_audio.append" (default: "session.input_audio.append") The event type, always session.input_audio.append.
OutputAudioDeltaEvent { delta, type, end_ms, start_ms } An audio chunk generated by the Live model. Decode and play primary WebSocket chunks in delivery order using the configured session audio format. Sideband connections receive reflected output audio with timestamps.
delta: string Base64-encoded raw audio. Primary WebSocket events use the session’s configured format; reflected sideband events use mono PCM16LE at 24 kHz.
type: "session.output_audio.delta" (default: "session.output_audio.delta") The event type, always session.output_audio.delta.
end_ms: optional integer Exclusive session-relative end in milliseconds. Required on reflected sideband events; omitted on the primary WebSocket. Dropped output frames leave gaps between reflected ranges.
start_ms: optional integer Inclusive session-relative start in milliseconds. Required on reflected sideband events; omitted on the primary WebSocket.
InputTranscriptDeltaEvent { delta, end_ms, event_id, 3 more } A transcript fragment for user input audio in the Live session. Accumulate fragments in delivery order; these events do not define complete turns or include a transcript-done event.
delta: string The transcript text fragment for the audio in this time range. Append fragments in delivery order to build the transcript.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.input_transcript.delta" (default: "session.input_transcript.delta") The event type, always session.input_transcript.delta.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
OutputTranscriptDeltaEvent { delta, end_ms, event_id, 3 more } A transcript fragment for assistant output audio in the Live session. Accumulate fragments in delivery order; these events do not define complete turns or include a transcript-done event.
delta: string The transcript text fragment for the audio in this time range. Append fragments in delivery order to build the transcript.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.output_transcript.delta" (default: "session.output_transcript.delta") The event type, always session.output_transcript.delta.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
DelegationCreatedEvent { delegation, event_id, offset_ms, 2 more } Returned when the Live model delegates work to your application, Responses, or Decisions. Contains delegation metadata and the position on the session timeline where the work was delegated.
delegation: { id, target, type, response_id } The delegated work identifier and destination. This object contains metadata, not the task text.
id: string The unique ID of the delegation. Use this as delegation_id when replying to client-owned work or correlating backend events.
target: "client" or "responses" or "decisions" Where the Live model delegated the work: client, responses, or decisions.
One of the following: "client" or "responses" or "decisions" Where the Live model delegated the work: client, responses, or decisions.
One of the following: "client" "responses" "decisions" type: "delegation" (default: "delegation") The object type, always delegation.
response_id: optional string The ID of the Responses API response associated with a Responses delegation. Omitted for client delegations.
event_id: string The unique ID of the Live server event.
offset_ms: integer The position on the Live session timeline where the delegation was created, in milliseconds from the beginning of the session.
type: "session.delegation.created" (default: "session.delegation.created") The event type, always session.delegation.created.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
DecisionEvent { delegation_id, event, event_id, 4 more } A Decision API event for a Live delegation or transcript trigger. For a trigger, source_event_id identifies the transcript delta that caused this evaluation.
delegation_id: string or null The Live delegation ID, or null for a transcript trigger.
event: { decision, type } decision: Decision { answers, model, usage } The native Decision API response, including its model, ordered answers, and token usage.
type: "decision.completed" (default: "decision.completed") event_id: string The unique ID of this Live server event.
source_event_id: string or null The event_id of the transcript delta that fired this trigger, or null for a delegation.
source_event_type: "session.input_transcript.delta" or "session.output_transcript.delta" or null One of the following: "session.input_transcript.delta" or "session.output_transcript.delta" One of the following: "session.input_transcript.delta" "session.output_transcript.delta" trigger_id: string or null The Live-assigned ID of this trigger firing, or null for a delegation.
type: "decision.event" (default: "decision.event") DecisionItemCreatedEvent { event_id, item, type, client_event_id } An input item was appended for future decisions.
event_id: string The unique ID of the Live server event.
item: Message or Message or Message or 5 more One of the following: Message { id, content, role, type } id: string content: string or array of InputText or InputImage or InputJSON The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of InputText or InputImage or InputJSON One of the following: InputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") role: "user" (default: "user") The message role. Always user.
type: optional "message" (default: "message") Message { id, content, role, type } id: string content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "developer" (default: "developer") type: optional "message" (default: "message") Message { id, content, role, type } id: string content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "system" (default: "system") type: optional "message" (default: "message") Message { id, content, role, 2 more } id: string content: string or array of OutputText or Refusal The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of OutputText or Refusal One of the following: OutputText { text, type } text: string The text content.
maxLength10485760 type: "output_text" (default: "output_text") Refusal { refusal, type } refusal: string The refusal text.
maxLength10485760 type: "refusal" (default: "refusal") role: "assistant" (default: "assistant") phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer). For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" type: optional "message" (default: "message") FunctionCall { id, arguments, call_id, 4 more } id: string arguments: string The function arguments as a JSON string.
call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 name: string pattern^[a-zA-Z0-9_.-]+$ minLength1 maxLength128 type: "function_call" (default: "function_call") The item type. Always function_call.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" FunctionCallOutput { id, call_id, output, 2 more } id: string call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "function_call_output" (default: "function_call_output") The type of the function tool call output. Always function_call_output.
status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCall { id, call_id, input, 4 more } id: string call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 input: string The input to the custom tool call.
name: string The name of the custom tool to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 type: "custom_tool_call" (default: "custom_tool_call") The item type. Always custom_tool_call.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCallOutput { id, call_id, output, 2 more } id: string call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "custom_tool_call_output" (default: "custom_tool_call_output") status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" type: "decision.item.created" (default: "decision.item.created") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
DecisionItemDeletedEvent { event_id, item_id, type, client_event_id } An input item was removed from future decisions.
event_id: string The unique ID of the Live server event.
item_id: string type: "decision.item.deleted" (default: "decision.item.deleted") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
ResponseEvent { event, event_id, type, 2 more } A streaming Responses API event from a backend delegated to by the Live session. Use the outer delegation_id to associate the nested stream with its Live delegation.
event: map[unknown] The nested Responses streaming event. Dispatch on its type field. Response lifecycle snapshots omit input and clear instructions, tools, and output to keep messages small; consume granular output events for the generated content.
event_id: string The unique ID of the Live server event.
type: "response.event" (default: "response.event") The event type, always response.event.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
delegation_id: optional string or null The Live delegation associated with the nested Responses event. May be null or omitted when the event cannot be correlated with a delegation.
SessionUsageUpdatedEvent { event_id, type, usage, 2 more } Reports cumulative Live audio usage and, when available, the most recent context-window usage. Delegated Responses token usage is reported separately in response.event events.
event_id: string The unique ID of the Live server event.
type: "session.usage.updated" (default: "session.usage.updated") The event type, always session.usage.updated.
usage: SessionUsage { seconds } The cumulative Live audio usage so far.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
context_window: optional { usage_ratio } The latest measured Live context-window usage. Omitted when the context limit is unknown.
usage_ratio: number The latest active context token count divided by the Live model context limit. Can decrease after compaction and may lag between measured audio frames.
SessionClosedEvent { event_id, reason, session, 3 more } Returned after the Live session finishes finalizing, with the close reason, final session snapshot, and cumulative audio usage. A connection closing without this event does not confirm successful finalization.
event_id: string The unique ID of the Live server event.
reason: "close_requested" or "expired" or "content" or 2 more Why the Live session ended: close_requested for an application close or hangup request, expired for the session duration limit, content for a safety filter, remote_hangup for a graceful remote disconnect, or connection_lost for an unexpected primary or upstream disconnection.
One of the following: "close_requested" or "expired" or "content" or 2 more Why the Live session ended: close_requested for an application close or hangup request, expired for the session duration limit, content for a safety filter, remote_hangup for a graceful remote disconnect, or connection_lost for an unexpected primary or upstream disconnection.
One of the following: "close_requested" "expired" "content" "remote_hangup" "connection_lost" session: SessionResource { id, expires_at, model, 9 more } The resolved Live session configuration and server-assigned session metadata.
type: "session.closed" (default: "session.closed") The event type, always session.closed.
usage: SessionUsage { seconds } The final cumulative Live audio usage after session finalization.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
ErrorEvent { error, event_id, type, client_event_id } Reports an error in the Live session, such as an invalid client command. Use error.client_event_id, when present, to identify the command that caused the error.
error: Error { code, message, type, 2 more } Details of the Live error and the client command that caused it, when known.
event_id: string The unique ID of the Live server event.
type: "error" (default: "error") The event type, always error.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InfoEvent { code, event_id, message, 2 more } An informational notice about the Live session, such as the event permissions applied to a frontend data channel.
code: string A machine-readable code for the notice, such as data_channel_permissions.
event_id: string The unique ID of the Live server event.
message: string A human-readable explanation of the Live session notice.
type: "info" (default: "info") The event type, always info.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
TransportDtmfReceived { event, event_id, type } A SIP DTMF keypress received from the caller. Delivered only to sideband observers.
event: string pattern^[0-9A-D*#]$ minLength1 maxLength1 event_id: string type: "transport.dtmf.received" (default: "transport.dtmf.received") TransportDtmfSent { event, event_id, type, client_event_id } A SIP DTMF keypress successfully sent to the SIP trunk. Delivered only to sideband observers; this is not a client command.
event: string pattern^[0-9A-D*#]$ minLength1 maxLength1 event_id: string type: "transport.dtmf.sent" (default: "transport.dtmf.sent") client_event_id: optional string The event_id of the client command, when supplied.
TransportRinging { event_id, session_id, type } The outbound SIP provider leg is ringing or providing early media. Delivered only to sideband observers.
event_id: string session_id: string The canonical Live session ID.
type: "transport.ringing" (default: "transport.ringing") TransportAnswered { event_id, session_id, type } The outbound SIP provider leg answered and media is established. Delivered only to sideband observers.
event_id: string session_id: string The canonical Live session ID.
type: "transport.answered" (default: "transport.answered") TransportFailed { error, event_id, session_id, type } An asynchronous outbound SIP setup failure. Delivered only to sideband observers.
error: { code, message, type, param } code: string The call setup failure code.
message: string type: "call_error" (default: "call_error") param: optional string The parameter related to the error, if any. Empty when no parameter applies.
event_id: string session_id: string The canonical Live session ID.
type: "transport.failed" (default: "transport.failed")
LiveSessions
Accept call
POST
/live/sessions/{session_id}/acceptDownload recording
GET
/live/sessions/{session_id}/contentFork session
POST
/live/sessions/{session_id}/forkHang up session
POST
/live/sessions/{session_id}/hangupTransfer call
POST
/live/sessions/{session_id}/referReject call
POST
/live/sessions/{session_id}/rejectModels
SessionForkResponse object { session, transport } The created Live session identifier and WebRTC answer. Apply transport.sdp as the peer’s remote answer and wait for session.started on the data channel before sending commands.
SessionForkResponse object { session, transport }
The created Live session identifier and WebRTC answer. Apply transport.sdp as the peer’s remote answer and wait for session.started on the data channel before sending commands.
session: { id } The newly created Live session. Use its ID for session controls and sideband connections.
id: string Opaque session identifier. Preserve the returned value unchanged, including its prefix.
transport: { sdp, type } WebRTC transport with the SDP answer.
sdp: string Session Description Protocol message for the WebRTC connection.
minLength1 type: "webrtc" The transport used for the Live session. Always webrtc.
LiveSideband
Connect
WebSocket
/live/sessions/{session_id}/attachModels
SidebandClientEvent = SessionUpdateEvent { session, type, event_id } or InputAudioMuteEvent { type, event_id } or InputAudioUnmuteEvent { type, event_id } or 11 moreClient events accepted by an attached Live sideband WebSocket. The session has already started. Send input audio over the primary connection.
SidebandClientEvent = SessionUpdateEvent { session, type, event_id } or InputAudioMuteEvent { type, event_id } or InputAudioUnmuteEvent { type, event_id } or 11 more
Client events accepted by an attached Live sideband WebSocket. The session has already started. Send input audio over the primary connection.
One of the following: SessionUpdateEvent { session, type, event_id } Update the delegation settings or transcript triggers of an active Live session. The server acknowledges accepted changes with session.updated.
session: SessionUpdateConfig { delegation, triggers } Sparse delegation and trigger updates. Omitted settings retain their values. The delegation type cannot change. Model, frontend instructions, audio, and startup input are immutable.
type: "session.update" (default: "session.update") The Live client event type. Always session.update.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 InputAudioMuteEvent { type, event_id } Mute audio input to the Live model without closing the session. The server acknowledges with session.input_audio.muted.
type: "session.input_audio.mute" (default: "session.input_audio.mute") The Live client event type. Always session.input_audio.mute.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 InputAudioUnmuteEvent { type, event_id } Resume audio input to a Live model after muting it. The server acknowledges with session.input_audio.unmuted. An active audio fixture still pauses inference until playback finishes.
type: "session.input_audio.unmute" (default: "session.input_audio.unmute") The Live client event type. Always session.input_audio.unmute.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 OutputAudioFixturePlayEvent { audio, type, event_id } Play one complete audio fixture through the session output. Send this event over the primary WebSocket or a server sideband. Audio must be raw mono PCM16 little-endian at 24 kHz, regardless of the session audio format, and at most 30 seconds. The entire payload is validated before model speech is interrupted and input is muted. Only one fixture can be active. fixture.playing acknowledges acceptance and fixture.played reports server mixer drain. These notifications are not exact transport-media boundaries. Input arriving during playback is dropped and not replayed.
audio: string Base64-encoded raw mono PCM16 audio, little-endian, at 24 kHz. Do not include a WAV or other container header.
minLength1 maxLength1920000 type: "session.output_audio.fixture.play" (default: "session.output_audio.fixture.play") event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 TransportDtmfSend { event, type, event_id } Send one DTMF keypress to the SIP trunk through a Live sideband connection. This command is supported only for SIP calls. A successful send emits transport.dtmf.sent on sideband. If you provide event_id, the success notification includes it as client_event_id; a failure includes it as error.client_event_id. Send separate events for multiple keypresses.
event: string The DTMF key to send: one of 0–9, A–D, *, or #.
pattern^[0-9A-D*#]$ minLength1 maxLength1 type: "transport.dtmf.send" (default: "transport.dtmf.send") event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 InstructionsAppendEvent { content, delegation_id, type, event_id } Append developer instructions to the Live conversation while it is running, optionally associating them with an existing client delegation. Rejected while an audio fixture is active. These instructions remain in conversation history through compaction and are included in future backend delegations. For external audio, describe content already delivered so the assistant does not repeat it.
content: string Instruction text to append, limited to 500 tokens. This is a plain string, not an array of content parts.
delegation_id: string or null Required, nullable. Set null for general session context, or use the ID from session.delegation.created for an existing client delegation. Non-null IDs are not accepted with Responses delegation.
minLength1 type: "session.instructions.append" (default: "session.instructions.append") The Live client event type. Always session.instructions.append.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 ThinkingAppendEvent { content, delegation_id, type, event_id } Provide silent reasoning or progress context to the Live model, optionally for an existing client delegation. Rejected while an audio fixture is active.
content: string Silent reasoning or progress context, limited to 500 tokens. It does not directly request speech, but can influence later speech and is not a secrecy boundary.
delegation_id: string or null Required, nullable. Set null for general session context, or use the ID from session.delegation.created for an existing client delegation. Non-null IDs are not accepted with Responses delegation.
minLength1 type: "session.thinking.append" (default: "session.thinking.append") The Live client event type. Always session.thinking.append.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 CommentaryAppendEvent { content, delegation_id, type, event_id } Provide context the Live model can communicate to the user, optionally for an existing client delegation. Rejected while an audio fixture is active.
content: string Speakable context for the Live model, limited to 500 tokens. Use this for a result the model should communicate; use session.thinking.append for silent context.
delegation_id: string or null Required, nullable. Set null for general session context, or use the ID from session.delegation.created for an existing client delegation. Non-null IDs are not accepted with Responses delegation.
minLength1 type: "session.commentary.append" (default: "session.commentary.append") The Live client event type. Always session.commentary.append.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 DecisionCreateEvent { event_id, type } Evaluate the configured questions using the current conversation and Decision inputs. Reusing an event_id is rejected for the lifetime of the session.
event_id: string minLength1 maxLength256 type: "decision.create" (default: "decision.create") DecisionItemCreateEvent { item, type, event_id } Append an input item for future decisions without starting an evaluation or changing Live voice context.
item: Message or Message or Message or 5 more One of the following: Message { content, role, id, type } content: string or array of InputText or InputImage or InputJSON The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of InputText or InputImage or InputJSON One of the following: InputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") role: "user" (default: "user") The message role. Always user.
id: optional string or null The unique ID of this message item.
type: optional "message" (default: "message") Message { content, role, id, type } content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "developer" (default: "developer") id: optional string or null The unique ID of this message item.
type: optional "message" (default: "message") Message { content, role, id, type } content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "system" (default: "system") id: optional string or null The unique ID of this message item.
type: optional "message" (default: "message") Message { content, role, id, 2 more } content: string or array of OutputText or Refusal The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of OutputText or Refusal One of the following: OutputText { text, type } text: string The text content.
maxLength10485760 type: "output_text" (default: "output_text") Refusal { refusal, type } refusal: string The refusal text.
maxLength10485760 type: "refusal" (default: "refusal") role: "assistant" (default: "assistant") id: optional string or null The unique ID of this message item.
phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer). For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" type: optional "message" (default: "message") FunctionCall { arguments, call_id, name, 4 more } arguments: string The function arguments as a JSON string.
call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 name: string pattern^[a-zA-Z0-9_.-]+$ minLength1 maxLength128 type: "function_call" (default: "function_call") The item type. Always function_call.
id: optional string or null The local Decision item ID.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" FunctionCallOutput { call_id, output, type, 2 more } call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "function_call_output" (default: "function_call_output") The type of the function tool call output. Always function_call_output.
id: optional string or null The local Decision item ID.
status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCall { call_id, input, name, 4 more } call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 input: string The input to the custom tool call.
name: string The name of the custom tool to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 type: "custom_tool_call" (default: "custom_tool_call") The item type. Always custom_tool_call.
id: optional string or null The local Decision item ID.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCallOutput { call_id, output, type, 2 more } call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "custom_tool_call_output" (default: "custom_tool_call_output") id: optional string or null The local Decision item ID.
status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" type: "decision.item.create" (default: "decision.item.create") event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 DecisionItemDeleteEvent { item_id, type, event_id } Delete an input item from future decisions without starting an evaluation or changing Live voice context.
item_id: string minLength1 type: "decision.item.delete" (default: "decision.item.delete") event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 ResponseItemCreateEvent { item, type, event_id } Add an input item to the Live session’s Responses backend. Requires Responses delegation; use response.create to request a response.
item: EasyInputMessage or Message or ResponseOutputMessage or 30 more An input item to append to the Responses backend conversation, such as a user message or a function tool result.
One of the following: EasyInputMessage { content, role, phase, type } A message input to the model with a role indicating instruction following
hierarchy. Instructions given with the developer or system role take
precedence over instructions given with the user role. Messages with the
assistant role are presumed to have been generated by the model in previous
interactions.
content: string or ResponseInputMessageContentList Text, image, or audio input to the model, used to generate a response.
Can also contain previous assistant responses.
One of the following: TextInput = string A text input to the model.
ResponseInputMessageContentList = array of ResponseInputContent A list of one or many input items to the model, containing different content
types.
One of the following: ResponseInputText { text, type, prompt_cache_breakpoint } A text input to the model.
text: string The text input to the model.
type: "input_text" (default: "input_text") The type of the input item. Always input_text.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputImage { detail, type, file_id, 2 more } An image input to the model. Learn about image inputs.
detail: ImageDetail The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
type: "input_image" (default: "input_image") The type of the input item. Always input_image.
file_id: optional string or null The ID of the file to be sent to the model.
image_url: optional string or null The URL of the image to be sent to the model. A fully qualified URL or base64 encoded image in a data URL.
formaturi prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputFile { type, detail, file_data, 4 more } A file input to the model.
type: "input_file" (default: "input_file") The type of the input item. Always input_file.
detail: optional "auto" or "low" or "high" The detail level of the file to be sent to the model. Use auto to let the system select the detail level; for GPT-5.6 and later models, auto uses high-quality rendering, which may increase input token usage. Use low for lower-cost rendering, or high to render the file at higher quality. Defaults to auto.
One of the following: "auto" "low" "high" file_data: optional string The content of the file to be sent to the model.
file_id: optional string or null The ID of the file to be sent to the model.
file_url: optional string The URL of the file to be sent to the model.
formaturi filename: optional string The name of the file to be sent to the model.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" role: "user" or "assistant" or "system" or "developer" The role of the message input. One of user, assistant, system, or
developer.
One of the following: "user" "assistant" "system" "developer" phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer).
For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend
phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" type: optional "message" The type of the message input. Always message.
Message { content, role, status, type } A message input to the model with a role indicating instruction following
hierarchy. Instructions given with the developer or system role take
precedence over instructions given with the user role.
content: ResponseInputMessageContentList { , , } A list of one or many input items to the model, containing different content
types.
role: "user" or "system" or "developer" The role of the message input. One of user, system, or developer.
One of the following: "user" "system" "developer" status: optional "in_progress" or "completed" or "incomplete" The status of item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" type: optional "message" The type of the message input. Always set to message.
ResponseOutputMessage { id, content, role, 3 more } An output message from the model.
id: string The unique ID of the output message.
content: array of ResponseOutputText or ResponseOutputRefusal The content of the output message.
One of the following: ResponseOutputText { annotations, logprobs, text, type } A text output from the model.
annotations: array of FileCitation or URLCitation or ContainerFileCitation or FilePath The annotations of the text output.
One of the following: FileCitation { file_id, filename, index, type } A citation to a file.
file_id: string The ID of the file.
filename: string The filename of the file cited.
index: integer The index in the output text at which to insert the file citation.
type: "file_citation" (default: "file_citation") The type of the file citation. Always file_citation.
URLCitation { end_index, start_index, title, 2 more } A citation for a web resource used to generate a model response.
end_index: integer The index of the last character of the URL citation in the message.
start_index: integer The index of the first character of the URL citation in the message.
title: string The title of the web resource.
type: "url_citation" (default: "url_citation") The type of the URL citation. Always url_citation.
url: string The URL of the web resource.
formaturi ContainerFileCitation { container_id, end_index, file_id, 3 more } A citation for a container file used to generate a model response.
container_id: string The ID of the container file.
end_index: integer The index of the last character of the container file citation in the message.
file_id: string The ID of the file.
filename: string The filename of the container file cited.
start_index: integer The index of the first character of the container file citation in the message.
type: "container_file_citation" (default: "container_file_citation") The type of the container file citation. Always container_file_citation.
FilePath { file_id, index, type } A path to a file.
file_id: string The ID of the file.
index: integer The index of the file in the list of files.
type: "file_path" The type of the file path. Always file_path.
logprobs: array of { token, bytes, logprob, top_logprobs } token: string bytes: array of number logprob: number top_logprobs: array of { token, bytes, logprob } token: string bytes: array of number logprob: number text: string The text output from the model.
type: "output_text" (default: "output_text") The type of the output text. Always output_text.
ResponseOutputRefusal { refusal, type } A refusal from the model.
refusal: string The refusal explanation from the model.
type: "refusal" (default: "refusal") The type of the refusal. Always refusal.
role: "assistant" The role of the output message. Always assistant.
status: "in_progress" or "completed" or "incomplete" The status of the message input. One of in_progress, completed, or
incomplete. Populated when input items are returned via API.
One of the following: "in_progress" "completed" "incomplete" type: "message" The type of the output message. Always message.
phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer).
For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend
phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" FileSearchCall { id, queries, status, 2 more } The results of a file search tool call. See the
file search guide for more information.
id: string The unique ID of the file search tool call.
queries: array of string The queries used to search for files.
status: "in_progress" or "searching" or "completed" or 2 more The status of the file search tool call. One of in_progress,
searching, incomplete or failed,
One of the following: "in_progress" "searching" "completed" "incomplete" "failed" type: "file_search_call" The type of the file search tool call. Always file_search_call.
results: optional array of object or null The results of the file search tool call.
attributes: optional map[string or number or boolean] or null Set of 16 key-value pairs that can be attached to an object. This can be
useful for storing additional information about the object in a structured
format, and querying for objects via API or the dashboard. Keys are strings
with a maximum length of 64 characters. Values are strings with a maximum
length of 512 characters, booleans, or numbers.
One of the following: string maxLength512 number boolean file_id: optional string The unique ID of the file.
filename: optional string The name of the file.
score: optional number The relevance score of the file - a value between 0 and 1.
formatfloat text: optional string The text that was retrieved from the file.
ComputerCall { id, call_id, pending_safety_checks, 4 more } A tool call to a computer use tool. See the
computer use guide for more information.
id: string The unique ID of the computer call.
call_id: string An identifier used when responding to the tool call with output.
pending_safety_checks: array of { id, code, message } The pending safety checks for the computer call.
id: string The ID of the pending safety check.
code: optional string or null The type of the pending safety check.
message: optional string or null Details about the pending safety check.
status: "in_progress" or "completed" or "incomplete" The status of the item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" type: "computer_call" (default: "computer_call") The type of the computer call. Always computer_call.
action: optional ComputerAction A click action.
actions: optional ComputerActionList { Click, DoubleClick, Drag, 6 more } Flattened batched actions for computer_use. Each action includes an
type discriminator and action-specific fields.
ComputerCallOutput { call_id, output, type, 3 more } The output of a computer tool call.
call_id: string The ID of the computer tool call that produced the output.
minLength1 maxLength64 output: ResponseComputerToolCallOutputScreenshot { type, file_id, image_url } A computer screenshot image used with the computer use tool.
type: "computer_call_output" (default: "computer_call_output") The type of the computer tool call output. Always computer_call_output.
id: optional string or null The ID of the computer tool call output.
acknowledged_safety_checks: optional array of object or null The safety checks reported by the API that have been acknowledged by the developer.
id: string The ID of the pending safety check.
code: optional string or null The type of the pending safety check.
message: optional string or null Details about the pending safety check.
status: optional "in_progress" or "completed" or "incomplete" or null The status of the message input. One of in_progress, completed, or incomplete. Populated when input items are returned via API.
One of the following: "in_progress" "completed" "incomplete" WebSearchCall { id, status, type, action } The results of a web search tool call. See the
web search guide for more information.
id: string The unique ID of the web search tool call.
status: "in_progress" or "searching" or "completed" or 2 more The status of the web search tool call.
One of the following: "in_progress" "searching" "completed" "failed" "incomplete" type: "web_search_call" The type of the web search tool call. Always web_search_call.
action: optional Search or OpenPage or FindInPage An object describing the specific action taken in this web search call.
Includes details on how the model used the web (search, open_page, find_in_page).
One of the following: Search { type, queries, query, sources } Action type “search” - Performs a web search query.
type: "search" The action type.
queries: optional array of string The search queries.
query: optional string Deprecated.
The search query.
sources: optional array of { type, url } The sources used in the search.
type: "url" The type of source. Always url.
url: string The URL of the source.
formaturi OpenPage { type, url } Action type “open_page” - Opens a specific URL from search results.
type: "open_page" The action type.
url: optional string or null The URL opened by the model.
formaturi FindInPage { pattern, type, url } Action type “find_in_page”: Searches for a pattern within a loaded page.
pattern: string The pattern or text to search for within the page.
type: "find_in_page" The action type.
url: optional string or null The URL of the page searched for the pattern, when available.
formaturi FunctionCall { arguments, call_id, name, 6 more } A tool call to run a function. See the
function calling guide for more information.
arguments: string A JSON string of the arguments to pass to the function.
call_id: string The unique ID of the function tool call generated by the model.
name: string The name of the function to run.
type: "function_call" The type of the function tool call. Always function_call.
id: optional string The unique ID of the function tool call.
async: optional boolean Whether the function tool call runs asynchronously.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
type: "program" (default: "program") namespace: optional string The namespace of the function to run.
status: optional "in_progress" or "completed" or "incomplete" The status of the item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" FunctionCallOutput { output, type, id, 5 more } The output of a function tool call.
output: string or array of ResponseInputTextContent or ResponseInputImageContent or ResponseInputFileContent Text, image, or file output of the function tool call.
One of the following: string A JSON string of the output of the function tool call.
maxLength10485760 array of ResponseInputTextContent or ResponseInputImageContent or ResponseInputFileContent An array of content outputs (text, image, file) for the function tool call.
One of the following: ResponseInputTextContent { text, type, prompt_cache_breakpoint } A text input to the model.
text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") The type of the input item. Always input_text.
prompt_cache_breakpoint: optional object or null Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputImageContent { type, detail, file_id, 2 more } An image input to the model. Learn about image inputs
type: "input_image" (default: "input_image") The type of the input item. Always input_image.
detail: optional ImageDetail or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
file_id: optional string or null The ID of the file to be sent to the model.
image_url: optional string or null The URL of the image to be sent to the model. A fully qualified URL or base64 encoded image in a data URL.
maxLength20971520 formaturi prompt_cache_breakpoint: optional object or null Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputFileContent { type, detail, file_data, 4 more } A file input to the model.
type: "input_file" (default: "input_file") The type of the input item. Always input_file.
detail: optional "auto" or "low" or "high" The detail level of the file to be sent to the model. Use auto to let the system select the detail level; for GPT-5.6 and later models, auto uses high-quality rendering, which may increase input token usage. Use low for lower-cost rendering, or high to render the file at higher quality. Defaults to auto.
One of the following: "auto" "low" "high" file_data: optional string or null The base64-encoded data of the file to be sent to the model.
maxLength73400320 file_id: optional string or null The ID of the file to be sent to the model.
file_url: optional string or null The URL of the file to be sent to the model.
formaturi filename: optional string or null The name of the file to be sent to the model.
prompt_cache_breakpoint: optional object or null Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" type: "function_call_output" (default: "function_call_output") The type of the function tool call output. Always function_call_output.
id: optional string or null The unique ID of the function tool call output. Populated when this item is returned via API.
call_id: optional string or null The unique ID of the function tool call generated by the model.
minLength1 maxLength64 caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
name: optional string or null The name of the tool that produced the output.
minLength1 maxLength128 namespace: optional string or null The namespace of the tool that produced the output.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 status: optional "in_progress" or "completed" or "incomplete" or null The status of the item. One of in_progress, completed, or incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" ToolSearchCall { arguments, type, id, 3 more } arguments: unknown The arguments supplied to the tool search call.
type: "tool_search_call" (default: "tool_search_call") The item type. Always tool_search_call.
id: optional string or null The unique ID of this tool search call.
call_id: optional string or null The unique ID of the tool search call generated by the model.
minLength1 maxLength64 execution: optional "server" or "client" Whether tool search was executed by the server or by the client.
One of the following: "server" "client" status: optional "in_progress" or "completed" or "incomplete" or null The status of the tool search call.
One of the following: "in_progress" "completed" "incomplete" ToolSearchOutput { tools, type, id, 3 more } tools: array of Function or FileSearch or Computer or 13 more The loaded tool definitions returned by the tool search output.
One of the following: Function { name, parameters, strict, 6 more } Defines a function in your own code the model can choose to call. Learn more about function calling.
name: string The name of the function to call.
parameters: map[unknown] or null A JSON schema object describing the parameters of the function.
strict: boolean or null Whether strict parameter validation is enforced for this function tool.
type: "function" (default: "function") The type of the function tool. Always function.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
One of the following: "direct" "programmatic" async: optional boolean defer_loading: optional boolean Whether this function is deferred and loaded via tool search.
description: optional string or null A description of the function. Used by the model to determine whether or not to call the function.
output_schema: optional map[unknown] or null A JSON schema object describing the JSON value encoded in string outputs for this function.
FileSearch { type, vector_store_ids, filters, 2 more } A tool that searches for relevant content from uploaded files. Learn more about the file search tool.
type: "file_search" (default: "file_search") The type of the file search tool. Always file_search.
vector_store_ids: array of string The IDs of the vector stores to search.
filters: optional ComparisonFilter or CompoundFilter or null A filter to apply.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } Combine multiple filters using and or or.
filters: array of ComparisonFilter or CompoundFilter Array of filters to combine. Items can be ComparisonFilter or CompoundFilter.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } This type repeats the enclosing schema. type: "and" or "or" Type of operation: and or or.
One of the following: "and" "or" max_num_results: optional integer The maximum number of results to return. This number should be between 1 and 50 inclusive.
ranking_options: optional { hybrid_search, ranker, score_threshold } Ranking options for search.
hybrid_search: optional { embedding_weight, text_weight } Weights that control how reciprocal rank fusion balances semantic embedding matches versus sparse keyword matches when hybrid search is enabled.
embedding_weight: number The weight of the embedding in the reciprocal ranking fusion.
text_weight: number The weight of the text in the reciprocal ranking fusion.
ranker: optional "auto" or "default-2024-11-15" The ranker to use for the file search.
One of the following: "auto" "default-2024-11-15" score_threshold: optional number The score threshold for the file search, a number between 0 and 1. Numbers closer to 1 will attempt to return only the most relevant results, but may return fewer results.
Computer { type } A tool that controls a virtual computer. Learn more about the computer tool.
type: "computer" (default: "computer") The type of the computer tool. Always computer.
ComputerUsePreview { display_height, display_width, environment, type } A tool that controls a virtual computer. Learn more about the computer tool.
display_height: integer The height of the computer display.
display_width: integer The width of the computer display.
environment: "windows" or "mac" or "linux" or 2 more The type of computer environment to control.
One of the following: "windows" "mac" "linux" "ubuntu" "browser" type: "computer_use_preview" (default: "computer_use_preview") The type of the computer use tool. Always computer_use_preview.
WebSearch { type, external_web_access, filters, 2 more } Search the Internet for sources related to the prompt. Learn more about the
web search tool.
type: "web_search" or "web_search_2025_08_26" (default: "web_search") The type of the web search tool. One of web_search or web_search_2025_08_26.
One of the following: "web_search" "web_search_2025_08_26" external_web_access: optional boolean (default: true) Allow live internet access for web search. Defaults to true when omitted. When false, the web search tool runs in offline/cache-only mode and will not fetch new external content.
filters: optional object or null Filters for the search.
allowed_domains: optional array of string or null (default: []) Allowed domains for the search. If not provided, all domains are allowed.
Subdomains of the provided domains are allowed as well.
Example: ["pubmed.ncbi.nlm.nih.gov"]
search_context_size: optional "low" or "medium" or "high" (default: "medium") High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the
United States. To avoid this fallback, pass {"type": "approximate"} without
location fields. To localize results, provide the relevant location fields.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
type: optional "approximate" (default: "approximate") The type of location approximation. Always approximate.
Mcp { server_label, type, allowed_callers, 9 more } Give the model access to additional tools via remote Model Context Protocol
(MCP) servers. Learn more about MCP.
server_label: string A label for this MCP server, used to identify it in tool calls.
type: "mcp" The type of the MCP tool. Always mcp.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" allowed_tools: optional array of string or McpToolFilter or null List of allowed tool names or a filter object.
One of the following: McpAllowedTools = array of string A string array of allowed tool names
McpToolFilter { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
authorization: optional string An OAuth access token that can be used with a remote MCP server, either
with a custom MCP server URL or a service connector. Your application
must handle the OAuth authorization flow and provide the token here.
connector_id: optional "connector_dropbox" or "connector_gmail" or "connector_googlecalendar" or 5 more Deprecated.
Identifier for service connectors, like those available in ChatGPT. One of
server_url, connector_id, or tunnel_id must be provided. Learn more
about service connectors here.
This field is deprecated for models released after September 1, 2026.
Use server_url to connect to a remote MCP server, or tunnel_id to
connect through a Secure MCP Tunnel.
Currently supported connector_id values are:
Dropbox: connector_dropbox
Gmail: connector_gmail
Google Calendar: connector_googlecalendar
Google Drive: connector_googledrive
Microsoft Teams: connector_microsoftteams
Outlook Calendar: connector_outlookcalendar
Outlook Email: connector_outlookemail
SharePoint: connector_sharepoint
One of the following: "connector_dropbox" "connector_gmail" "connector_googlecalendar" "connector_googledrive" "connector_microsoftteams" "connector_outlookcalendar" "connector_outlookemail" "connector_sharepoint" defer_loading: optional boolean Whether this MCP tool is deferred and discovered via tool search.
headers: optional map[string] or null Optional HTTP headers to send to the MCP server. Use for authentication
or other purposes.
require_approval: optional McpToolApprovalFilter or "always" or "never" or null (default: "always") Specify which of the MCP server’s tools require approval.
One of the following: McpToolApprovalFilter { always, never } Specify which of the MCP server’s tools require approval. Can be
always, never, or a filter object associated with tools
that require approval.
always: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
never: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
McpToolApprovalSetting = "always" or "never" Specify a single approval policy for all tools. One of always or
never. When set to always, all tools will require approval. When
set to never, all tools will not require approval.
One of the following: "always" "never" server_description: optional string Optional description of the MCP server, used to provide more context.
server_url: optional string The URL for the MCP server. One of server_url, connector_id, or
tunnel_id must be provided.
formaturi tunnel_id: optional string The Secure MCP Tunnel ID to use instead of a direct server URL. One of
server_url, connector_id, or tunnel_id must be provided.
pattern^tunnel_[a-z0-9]{32}$ CodeInterpreter { container, type, allowed_callers } A tool that runs Python code to help generate a response to a prompt.
container: string or CodeInterpreterToolAuto The code interpreter container. Can be a container ID or an object that
specifies uploaded file IDs to make available to your code, along with an
optional memory_limit setting.
One of the following: string The container ID.
CodeInterpreterToolAuto { type, file_ids, memory_limit, network_policy } Configuration for a code interpreter container. Optionally specify the IDs of the files to run the code on.
type: "auto" (default: "auto") Always auto.
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the code interpreter container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 type: "code_interpreter" The type of the code interpreter tool. Always code_interpreter.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" ProgrammaticToolCalling { type } type: "programmatic_tool_calling" (default: "programmatic_tool_calling") The type of the tool. Always programmatic_tool_calling.
ImageGeneration { type, action, background, 13 more } A tool that generates images using the GPT image models.
type: "image_generation" The type of the image generation tool. Always image_generation.
action: optional "generate" or "edit" or "auto" Whether to generate a new image or edit an existing image. Default: auto.
One of the following: "generate" "edit" "auto" background: optional "transparent" or "opaque" or "auto" (default: "auto") Set the background of the generated image. One of transparent, opaque,
or auto. gpt-image-2.5-sunburst and gpt-image-2.5-flare, including
their 2026-09-08 snapshots, support opaque and transparent
backgrounds. Transparent backgrounds are available for supported GPT Image
models. For gpt-image-2 and gpt-image-2-2026-04-21, this support is in
preview. When using transparent, set the output format to png or webp.
Default: auto.
One of the following: "transparent" "opaque" "auto" complexity: optional "low" or "medium" or "high" Computation spent planning the image, independently of rendering quality.
Overrides the outer response’s reasoning.effort for image generation.
When omitted, inherits a supported outer reasoning effort or uses the
image model’s configured default.
For SVG output, controls the reference image, not the SVG author’s
reasoning.
One of the following: "low" "medium" "high" frame_count: optional integer (default: 9) Number of GIF frames. Requires output_format to be gif. Default: 9.
minimum2 maximum16 frame_duration_ms: optional integer (default: 200) Duration of each GIF frame in milliseconds, rounded to 10 ms.
Requires output_format to be gif. Default: 200.
minimum20 maximum2000 input_fidelity: optional "high" or "low" or null Control how much effort the model will exert to match the style and features, especially facial features, of input images. Supports high and low on gpt-image-1.5; gpt-image-1-mini supports only low. For gpt-image-2, omit this parameter. Defaults to low on supported models.
One of the following: "high" "low" input_image_mask: optional { file_id, image_url } Optional mask for inpainting. Contains image_url
(string, optional) and file_id (string, optional).
file_id: optional string File ID for the mask image.
image_url: optional string Base64-encoded mask image.
model: optional string or "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: string "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: "gpt-image-1-mini" "gpt-image-2" "gpt-image-2-2026-04-21" "gpt-image-2.5-sunburst" "gpt-image-2.5-sunburst-2026-09-08" "gpt-image-2.5-flare" "gpt-image-2.5-flare-2026-09-08" "gpt-image-2.6-sunburst" "gpt-image-2.6-flare" "gpt-image-1.5" "chatgpt-image-latest" "gpt-image-1" Deprecated. GPT Image 1 was shut down on October 23, 2026. Use gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
moderation: optional "auto" or "low" (default: "auto") Moderation level for the generated image. Default: auto.
One of the following: "auto" "low" output_compression: optional integer (default: 100) Compression level for the output image. Default: 100.
minimum0 maximum100 output_format: optional "png" or "webp" or "jpeg" or 2 more (default: "png") The output format of the generated image. One of png, webp, jpeg,
gif, or svg. Default: png. GIF and SVG output require a model
that supports the selected format. GIF output does not support
partial image streaming (partial_images greater than 0),
input_image_mask, or custom output_compression.
Responses lifecycle events can still be streamed for GIF output.
One of the following: "png" "webp" "jpeg" "gif" "svg" partial_images: optional integer (default: 0) Number of partial images to generate in streaming mode, from 0 (default value) to 3.
minimum0 maximum3 quality: optional "low" or "medium" or "high" or 3 more (default: "auto") The quality of the generated image. The GPT image models support low,
medium, and high. gpt-image-2.5-sunburst and gpt-image-2.5-flare,
including their 2026-09-08 snapshots, also support xhigh and max.
Default: auto.
One of the following: "low" "medium" "high" "xhigh" "max" "auto" size: optional string or "1024x1024" or "1024x1536" or "1536x1024" or "auto" (default: "auto") The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: string "1024x1024" or "1024x1536" or "1536x1024" or "auto" The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: "1024x1024" "1024x1536" "1536x1024" "auto" svg: optional { model, reasoning } SVG author model settings. Requires output_format to be svg.
When SVG model selection is enabled, omitting this object uses the
outer response model. Otherwise, the default SVG author is used and
explicit model overrides are unavailable.
model: optional string minLength1 reasoning: optional object or null Reasoning effort for SVG authoring. When omitted, uses the author model’s default in both Images and Responses.
effort: optional "none" or "minimal" or "low" or 4 more Controls how much effort the model should spend on reasoning for this request.
One of the following: "none" "minimal" "low" "medium" "high" "xhigh" "max" LocalShell { type } A tool that allows the model to execute shell commands in a local environment.
type: "local_shell" (default: "local_shell") The type of the local shell tool. Always local_shell.
Shell { type, allowed_callers, environment } A tool that allows the model to execute shell commands.
type: "shell" (default: "shell") The type of the shell tool. Always shell.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" environment: optional ContainerAuto or LocalEnvironment or ContainerReference or null One of the following: ContainerAuto { type, file_ids, memory_limit, 2 more } type: "container_auto" (default: "container_auto") Automatically creates a container for this request
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 skills: optional array of SkillReference or InlineSkill An optional list of skills referenced by id or inline data.
maxItems200 One of the following: SkillReference { skill_id, type, version } skill_id: string The ID of the referenced skill.
minLength1 maxLength64 type: "skill_reference" (default: "skill_reference") References a skill created with the /v1/skills endpoint.
version: optional string Optional skill version. Use a positive integer or ‘latest’. Omit for default.
InlineSkill { description, name, source, type } description: string The description of the skill.
name: string The name of the skill.
source: InlineSkillSource { data, media_type, type } Inline skill payload
type: "inline" (default: "inline") Defines an inline skill for this request.
LocalEnvironment { type, skills } type: "local" (default: "local") Use a local computer environment.
skills: optional array of LocalSkill { description, name, path } An optional list of skills.
maxItems200 description: string The description of the skill.
name: string The name of the skill.
path: string The path to the directory containing the skill.
ContainerReference { container_id, type } container_id: string The ID of the referenced container.
type: "container_reference" (default: "container_reference") References a container created with the /v1/containers endpoint
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
ToolSearchOutputNamespaceTool { description, name, tools, type } Groups function/custom tools under a shared namespace.
description: string A description of the namespace shown to the model.
name: string The namespace name used in tool calls (for example, crm).
minLength1 tools: array of Function or Custom The function/custom tools loaded inside this namespace.
minItems1 One of the following: Function { name, type, allowed_callers, 6 more } name: string The name of the loaded function tool.
pattern^(?![\s\S]*[\r\n])[a-zA-Z0-9_.-]+$ minLength1 maxLength128 type: "function" (default: "function") allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this function should be deferred and discovered via tool search.
description: optional string or null output_schema: optional map[unknown] or null A JSON Schema describing the JSON value encoded in string outputs for this function tool. This does not describe content-array outputs.
parameters: optional unknown or null strict: optional boolean or null Whether to enforce strict parameter validation. If omitted, Responses attempts to use strict validation when the schema is compatible, and falls back to non-strict validation otherwise.
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
type: "namespace" (default: "namespace") The type of the tool. Always namespace.
ToolSearch { type, description, execution, parameters } Hosted or BYOT tool search configuration for deferred tools.
type: "tool_search" (default: "tool_search") The type of the tool. Always tool_search.
description: optional string or null Description shown to the model for a client-executed tool search tool.
execution: optional "server" or "client" Whether tool search is executed by the server or by the client.
One of the following: "server" "client" parameters: optional unknown or null Parameter schema for a client-executed tool search tool.
WebSearchPreview { type, search_content_types, search_context_size, user_location } This tool searches the web for relevant results to use in a response. Learn more about the web search tool.
type: "web_search_preview" or "web_search_preview_2025_03_11" (default: "web_search_preview") The type of the web search tool. One of web_search_preview or web_search_preview_2025_03_11.
One of the following: "web_search_preview" "web_search_preview_2025_03_11" search_content_types: optional array of "text" or "image" One of the following: "text" "image" search_context_size: optional "low" or "medium" or "high" High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the United States. To avoid this fallback, pass {"type": "approximate"} without location fields. To localize results, provide the relevant location fields.
type: "approximate" (default: "approximate") The type of location approximation. Always approximate.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
ApplyPatch { type, allowed_callers } Allows the assistant to create, delete, or update files using unified diffs.
type: "apply_patch" (default: "apply_patch") The type of the tool. Always apply_patch.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" type: "tool_search_output" (default: "tool_search_output") The item type. Always tool_search_output.
id: optional string or null The unique ID of this tool search output.
call_id: optional string or null The unique ID of the tool search call generated by the model.
minLength1 maxLength64 execution: optional "server" or "client" Whether tool search was executed by the server or by the client.
One of the following: "server" "client" status: optional "in_progress" or "completed" or "incomplete" or null The status of the tool search output.
One of the following: "in_progress" "completed" "incomplete" AdditionalTools { role, tools, type, id } role: "developer" (default: "developer") The role that provided the additional tools. Only developer is supported.
tools: array of Function or FileSearch or Computer or 13 more A list of additional tools made available at this item.
One of the following: Function { name, parameters, strict, 6 more } Defines a function in your own code the model can choose to call. Learn more about function calling.
name: string The name of the function to call.
parameters: map[unknown] or null A JSON schema object describing the parameters of the function.
strict: boolean or null Whether strict parameter validation is enforced for this function tool.
type: "function" (default: "function") The type of the function tool. Always function.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
One of the following: "direct" "programmatic" async: optional boolean defer_loading: optional boolean Whether this function is deferred and loaded via tool search.
description: optional string or null A description of the function. Used by the model to determine whether or not to call the function.
output_schema: optional map[unknown] or null A JSON schema object describing the JSON value encoded in string outputs for this function.
FileSearch { type, vector_store_ids, filters, 2 more } A tool that searches for relevant content from uploaded files. Learn more about the file search tool.
type: "file_search" (default: "file_search") The type of the file search tool. Always file_search.
vector_store_ids: array of string The IDs of the vector stores to search.
filters: optional ComparisonFilter or CompoundFilter or null A filter to apply.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } Combine multiple filters using and or or.
filters: array of ComparisonFilter or CompoundFilter Array of filters to combine. Items can be ComparisonFilter or CompoundFilter.
One of the following: ComparisonFilter { key, type, value } A filter used to compare a specified attribute key to a given value using a defined comparison operation.
key: string The key to compare against the value.
type: "eq" or "ne" or "gt" or 5 more (default: "eq") Specifies the comparison operator: eq, ne, gt, gte, lt, lte, in, nin.
eq: equals
ne: not equal
gt: greater than
gte: greater than or equal
lt: less than
lte: less than or equal
in: in
nin: not in
One of the following: "eq" "ne" "gt" "gte" "lt" "lte" "in" "nin" value: string or number or boolean or array of string or number The value to compare against the attribute key; supports string, number, or boolean types.
One of the following: string number boolean array of string or number One of the following: string number CompoundFilter { filters, type } This type repeats the enclosing schema. type: "and" or "or" Type of operation: and or or.
One of the following: "and" "or" max_num_results: optional integer The maximum number of results to return. This number should be between 1 and 50 inclusive.
ranking_options: optional { hybrid_search, ranker, score_threshold } Ranking options for search.
hybrid_search: optional { embedding_weight, text_weight } Weights that control how reciprocal rank fusion balances semantic embedding matches versus sparse keyword matches when hybrid search is enabled.
embedding_weight: number The weight of the embedding in the reciprocal ranking fusion.
text_weight: number The weight of the text in the reciprocal ranking fusion.
ranker: optional "auto" or "default-2024-11-15" The ranker to use for the file search.
One of the following: "auto" "default-2024-11-15" score_threshold: optional number The score threshold for the file search, a number between 0 and 1. Numbers closer to 1 will attempt to return only the most relevant results, but may return fewer results.
Computer { type } A tool that controls a virtual computer. Learn more about the computer tool.
type: "computer" (default: "computer") The type of the computer tool. Always computer.
ComputerUsePreview { display_height, display_width, environment, type } A tool that controls a virtual computer. Learn more about the computer tool.
display_height: integer The height of the computer display.
display_width: integer The width of the computer display.
environment: "windows" or "mac" or "linux" or 2 more The type of computer environment to control.
One of the following: "windows" "mac" "linux" "ubuntu" "browser" type: "computer_use_preview" (default: "computer_use_preview") The type of the computer use tool. Always computer_use_preview.
WebSearch { type, external_web_access, filters, 2 more } Search the Internet for sources related to the prompt. Learn more about the
web search tool.
type: "web_search" or "web_search_2025_08_26" (default: "web_search") The type of the web search tool. One of web_search or web_search_2025_08_26.
One of the following: "web_search" "web_search_2025_08_26" external_web_access: optional boolean (default: true) Allow live internet access for web search. Defaults to true when omitted. When false, the web search tool runs in offline/cache-only mode and will not fetch new external content.
filters: optional object or null Filters for the search.
allowed_domains: optional array of string or null (default: []) Allowed domains for the search. If not provided, all domains are allowed.
Subdomains of the provided domains are allowed as well.
Example: ["pubmed.ncbi.nlm.nih.gov"]
search_context_size: optional "low" or "medium" or "high" (default: "medium") High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the
United States. To avoid this fallback, pass {"type": "approximate"} without
location fields. To localize results, provide the relevant location fields.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
type: optional "approximate" (default: "approximate") The type of location approximation. Always approximate.
Mcp { server_label, type, allowed_callers, 9 more } Give the model access to additional tools via remote Model Context Protocol
(MCP) servers. Learn more about MCP.
server_label: string A label for this MCP server, used to identify it in tool calls.
type: "mcp" The type of the MCP tool. Always mcp.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" allowed_tools: optional array of string or McpToolFilter or null List of allowed tool names or a filter object.
One of the following: McpAllowedTools = array of string A string array of allowed tool names
McpToolFilter { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
authorization: optional string An OAuth access token that can be used with a remote MCP server, either
with a custom MCP server URL or a service connector. Your application
must handle the OAuth authorization flow and provide the token here.
connector_id: optional "connector_dropbox" or "connector_gmail" or "connector_googlecalendar" or 5 more Deprecated.
Identifier for service connectors, like those available in ChatGPT. One of
server_url, connector_id, or tunnel_id must be provided. Learn more
about service connectors here.
This field is deprecated for models released after September 1, 2026.
Use server_url to connect to a remote MCP server, or tunnel_id to
connect through a Secure MCP Tunnel.
Currently supported connector_id values are:
Dropbox: connector_dropbox
Gmail: connector_gmail
Google Calendar: connector_googlecalendar
Google Drive: connector_googledrive
Microsoft Teams: connector_microsoftteams
Outlook Calendar: connector_outlookcalendar
Outlook Email: connector_outlookemail
SharePoint: connector_sharepoint
One of the following: "connector_dropbox" "connector_gmail" "connector_googlecalendar" "connector_googledrive" "connector_microsoftteams" "connector_outlookcalendar" "connector_outlookemail" "connector_sharepoint" defer_loading: optional boolean Whether this MCP tool is deferred and discovered via tool search.
headers: optional map[string] or null Optional HTTP headers to send to the MCP server. Use for authentication
or other purposes.
require_approval: optional McpToolApprovalFilter or "always" or "never" or null (default: "always") Specify which of the MCP server’s tools require approval.
One of the following: McpToolApprovalFilter { always, never } Specify which of the MCP server’s tools require approval. Can be
always, never, or a filter object associated with tools
that require approval.
always: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
never: optional { read_only, tool_names } A filter object to specify which tools are allowed.
read_only: optional boolean Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
tool_names: optional array of string List of allowed tool names.
McpToolApprovalSetting = "always" or "never" Specify a single approval policy for all tools. One of always or
never. When set to always, all tools will require approval. When
set to never, all tools will not require approval.
One of the following: "always" "never" server_description: optional string Optional description of the MCP server, used to provide more context.
server_url: optional string The URL for the MCP server. One of server_url, connector_id, or
tunnel_id must be provided.
formaturi tunnel_id: optional string The Secure MCP Tunnel ID to use instead of a direct server URL. One of
server_url, connector_id, or tunnel_id must be provided.
pattern^tunnel_[a-z0-9]{32}$ CodeInterpreter { container, type, allowed_callers } A tool that runs Python code to help generate a response to a prompt.
container: string or CodeInterpreterToolAuto The code interpreter container. Can be a container ID or an object that
specifies uploaded file IDs to make available to your code, along with an
optional memory_limit setting.
One of the following: string The container ID.
CodeInterpreterToolAuto { type, file_ids, memory_limit, network_policy } Configuration for a code interpreter container. Optionally specify the IDs of the files to run the code on.
type: "auto" (default: "auto") Always auto.
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the code interpreter container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 type: "code_interpreter" The type of the code interpreter tool. Always code_interpreter.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" ProgrammaticToolCalling { type } type: "programmatic_tool_calling" (default: "programmatic_tool_calling") The type of the tool. Always programmatic_tool_calling.
ImageGeneration { type, action, background, 13 more } A tool that generates images using the GPT image models.
type: "image_generation" The type of the image generation tool. Always image_generation.
action: optional "generate" or "edit" or "auto" Whether to generate a new image or edit an existing image. Default: auto.
One of the following: "generate" "edit" "auto" background: optional "transparent" or "opaque" or "auto" (default: "auto") Set the background of the generated image. One of transparent, opaque,
or auto. gpt-image-2.5-sunburst and gpt-image-2.5-flare, including
their 2026-09-08 snapshots, support opaque and transparent
backgrounds. Transparent backgrounds are available for supported GPT Image
models. For gpt-image-2 and gpt-image-2-2026-04-21, this support is in
preview. When using transparent, set the output format to png or webp.
Default: auto.
One of the following: "transparent" "opaque" "auto" complexity: optional "low" or "medium" or "high" Computation spent planning the image, independently of rendering quality.
Overrides the outer response’s reasoning.effort for image generation.
When omitted, inherits a supported outer reasoning effort or uses the
image model’s configured default.
For SVG output, controls the reference image, not the SVG author’s
reasoning.
One of the following: "low" "medium" "high" frame_count: optional integer (default: 9) Number of GIF frames. Requires output_format to be gif. Default: 9.
minimum2 maximum16 frame_duration_ms: optional integer (default: 200) Duration of each GIF frame in milliseconds, rounded to 10 ms.
Requires output_format to be gif. Default: 200.
minimum20 maximum2000 input_fidelity: optional "high" or "low" or null Control how much effort the model will exert to match the style and features, especially facial features, of input images. Supports high and low on gpt-image-1.5; gpt-image-1-mini supports only low. For gpt-image-2, omit this parameter. Defaults to low on supported models.
One of the following: "high" "low" input_image_mask: optional { file_id, image_url } Optional mask for inpainting. Contains image_url
(string, optional) and file_id (string, optional).
file_id: optional string File ID for the mask image.
image_url: optional string Base64-encoded mask image.
model: optional string or "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: string "gpt-image-1" or "gpt-image-1-mini" or "gpt-image-2" or 9 more The image generation model to use. One of gpt-image-1-mini,
gpt-image-1.5, gpt-image-2,
gpt-image-2-2026-04-21, gpt-image-2.5-sunburst,
gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare,
gpt-image-2.5-flare-2026-09-08, gpt-image-2.6-sunburst,
gpt-image-2.6-flare, or chatgpt-image-latest. Default:
gpt-image-2.6-sunburst.
gpt-image-1 was shut down on October 23, 2026. Use
gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
One of the following: "gpt-image-1-mini" "gpt-image-2" "gpt-image-2-2026-04-21" "gpt-image-2.5-sunburst" "gpt-image-2.5-sunburst-2026-09-08" "gpt-image-2.5-flare" "gpt-image-2.5-flare-2026-09-08" "gpt-image-2.6-sunburst" "gpt-image-2.6-flare" "gpt-image-1.5" "chatgpt-image-latest" "gpt-image-1" Deprecated. GPT Image 1 was shut down on October 23, 2026. Use gpt-image-2.5-sunburst or gpt-image-2.5-flare instead.
moderation: optional "auto" or "low" (default: "auto") Moderation level for the generated image. Default: auto.
One of the following: "auto" "low" output_compression: optional integer (default: 100) Compression level for the output image. Default: 100.
minimum0 maximum100 output_format: optional "png" or "webp" or "jpeg" or 2 more (default: "png") The output format of the generated image. One of png, webp, jpeg,
gif, or svg. Default: png. GIF and SVG output require a model
that supports the selected format. GIF output does not support
partial image streaming (partial_images greater than 0),
input_image_mask, or custom output_compression.
Responses lifecycle events can still be streamed for GIF output.
One of the following: "png" "webp" "jpeg" "gif" "svg" partial_images: optional integer (default: 0) Number of partial images to generate in streaming mode, from 0 (default value) to 3.
minimum0 maximum3 quality: optional "low" or "medium" or "high" or 3 more (default: "auto") The quality of the generated image. The GPT image models support low,
medium, and high. gpt-image-2.5-sunburst and gpt-image-2.5-flare,
including their 2026-09-08 snapshots, also support xhigh and max.
Default: auto.
One of the following: "low" "medium" "high" "xhigh" "max" "auto" size: optional string or "1024x1024" or "1024x1536" or "1536x1024" or "auto" (default: "auto") The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: string "1024x1024" or "1024x1536" or "1536x1024" or "auto" The size of the generated images. For gpt-image-2, gpt-image-2-2026-04-21, gpt-image-2.5-sunburst, gpt-image-2.5-sunburst-2026-09-08, gpt-image-2.5-flare, and gpt-image-2.5-flare-2026-09-08, arbitrary resolutions are supported as WIDTHxHEIGHT strings, for example 1536x864. Width and height must both be divisible by 16 and the requested aspect ratio must be between 1:3 and 3:1. Resolutions above 2560x1440 are experimental, and the maximum supported resolution is 3840x2160. The requested size must also satisfy the model’s current pixel and edge limits. The standard sizes 1024x1024, 1536x1024, and 1024x1536 are supported by the GPT image models; auto is supported for models that allow automatic sizing.
One of the following: "1024x1024" "1024x1536" "1536x1024" "auto" svg: optional { model, reasoning } SVG author model settings. Requires output_format to be svg.
When SVG model selection is enabled, omitting this object uses the
outer response model. Otherwise, the default SVG author is used and
explicit model overrides are unavailable.
model: optional string minLength1 reasoning: optional object or null Reasoning effort for SVG authoring. When omitted, uses the author model’s default in both Images and Responses.
effort: optional "none" or "minimal" or "low" or 4 more Controls how much effort the model should spend on reasoning for this request.
One of the following: "none" "minimal" "low" "medium" "high" "xhigh" "max" LocalShell { type } A tool that allows the model to execute shell commands in a local environment.
type: "local_shell" (default: "local_shell") The type of the local shell tool. Always local_shell.
Shell { type, allowed_callers, environment } A tool that allows the model to execute shell commands.
type: "shell" (default: "shell") The type of the shell tool. Always shell.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" environment: optional ContainerAuto or LocalEnvironment or ContainerReference or null One of the following: ContainerAuto { type, file_ids, memory_limit, 2 more } type: "container_auto" (default: "container_auto") Automatically creates a container for this request
file_ids: optional array of string An optional list of uploaded files to make available to your code.
maxItems50 memory_limit: optional "1g" or "4g" or "16g" or "64g" or null The memory limit for the container.
One of the following: "1g" "4g" "16g" "64g" network_policy: optional ContainerNetworkPolicyDisabled or ContainerNetworkPolicyAllowlist Network access policy for the container.
One of the following: ContainerNetworkPolicyDisabled { type } type: "disabled" (default: "disabled") Disable outbound network access. Always disabled.
ContainerNetworkPolicyAllowlist { allowed_domains, type, domain_secrets } allowed_domains: array of string A list of allowed domains when type is allowlist.
minItems1 type: "allowlist" (default: "allowlist") Allow outbound network access only to specified domains. Always allowlist.
domain_secrets: optional array of ContainerNetworkPolicyDomainSecret { domain, name, value } Optional domain-scoped secrets for allowlisted domains.
minItems1 domain: string The domain associated with the secret.
minLength1 name: string The name of the secret to inject for the domain.
minLength1 value: string The secret value to inject for the domain.
minLength1 maxLength10485760 skills: optional array of SkillReference or InlineSkill An optional list of skills referenced by id or inline data.
maxItems200 One of the following: SkillReference { skill_id, type, version } skill_id: string The ID of the referenced skill.
minLength1 maxLength64 type: "skill_reference" (default: "skill_reference") References a skill created with the /v1/skills endpoint.
version: optional string Optional skill version. Use a positive integer or ‘latest’. Omit for default.
InlineSkill { description, name, source, type } description: string The description of the skill.
name: string The name of the skill.
source: InlineSkillSource { data, media_type, type } Inline skill payload
type: "inline" (default: "inline") Defines an inline skill for this request.
LocalEnvironment { type, skills } type: "local" (default: "local") Use a local computer environment.
skills: optional array of LocalSkill { description, name, path } An optional list of skills.
maxItems200 description: string The description of the skill.
name: string The name of the skill.
path: string The path to the directory containing the skill.
ContainerReference { container_id, type } container_id: string The ID of the referenced container.
type: "container_reference" (default: "container_reference") References a container created with the /v1/containers endpoint
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
Namespace { description, name, tools, type } Groups function/custom tools under a shared namespace.
description: string A description of the namespace shown to the model.
name: string The namespace name used in tool calls (for example, crm).
minLength1 tools: array of Function or Custom The function/custom tools available inside this namespace.
minItems1 One of the following: Function { name, type, allowed_callers, 6 more } name: string pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 type: "function" (default: "function") allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this function should be deferred and discovered via tool search.
description: optional string or null output_schema: optional map[unknown] or null A JSON Schema describing the JSON value encoded in string outputs for this function tool. This does not describe content-array outputs.
parameters: optional unknown or null strict: optional boolean or null Whether to enforce strict parameter validation. If omitted, Responses attempts to use strict validation when the schema is compatible, and falls back to non-strict validation otherwise.
Custom { name, type, allowed_callers, 4 more } A custom tool that processes input using a specified format. Learn more about custom tools
name: string The name of the custom tool, used to identify it in tool calls.
type: "custom" (default: "custom") The type of the custom tool. Always custom.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" async: optional boolean Whether the tool response can be returned asynchronously versus immediately returned on next response creation.
defer_loading: optional boolean Whether this tool should be deferred and discovered via tool search.
description: optional string Optional description of the custom tool, used to provide more context.
format: optional CustomToolInputFormat The input format for the custom tool. Default is unconstrained text.
type: "namespace" (default: "namespace") The type of the tool. Always namespace.
ToolSearch { type, description, execution, parameters } Hosted or BYOT tool search configuration for deferred tools.
type: "tool_search" (default: "tool_search") The type of the tool. Always tool_search.
description: optional string or null Description shown to the model for a client-executed tool search tool.
execution: optional "server" or "client" Whether tool search is executed by the server or by the client.
One of the following: "server" "client" parameters: optional unknown or null Parameter schema for a client-executed tool search tool.
WebSearchPreview { type, search_content_types, search_context_size, user_location } This tool searches the web for relevant results to use in a response. Learn more about the web search tool.
type: "web_search_preview" or "web_search_preview_2025_03_11" (default: "web_search_preview") The type of the web search tool. One of web_search_preview or web_search_preview_2025_03_11.
One of the following: "web_search_preview" "web_search_preview_2025_03_11" search_content_types: optional array of "text" or "image" One of the following: "text" "image" search_context_size: optional "low" or "medium" or "high" High level guidance for the amount of context window space to use for the search. One of low, medium, or high. medium is the default.
One of the following: "low" "medium" "high" user_location: optional object or null The approximate location of the user. If omitted or null, defaults to the United States. To avoid this fallback, pass {"type": "approximate"} without location fields. To localize results, provide the relevant location fields.
type: "approximate" (default: "approximate") The type of location approximation. Always approximate.
city: optional string or null Free text input for the city of the user, e.g. San Francisco.
country: optional string or null The two-letter ISO country code of the user, e.g. US.
region: optional string or null Free text input for the region of the user, e.g. California.
timezone: optional string or null The IANA timezone of the user, e.g. America/Los_Angeles.
ApplyPatch { type, allowed_callers } Allows the assistant to create, delete, or update files using unified diffs.
type: "apply_patch" (default: "apply_patch") The type of the tool. Always apply_patch.
allowed_callers: optional array of "direct" or "programmatic" or null The tool invocation context(s).
minItems1 One of the following: "direct" "programmatic" type: "additional_tools" (default: "additional_tools") The item type. Always additional_tools.
id: optional string or null The unique ID of this additional tools item.
ConfigurationUpdate { type, id, reasoning, text } An update to the conversation’s response configuration. Supply reasoning,
text, or both. Each setting stays in effect until a later update replaces it.
type: "configuration_update" (default: "configuration_update") The item type. Always configuration_update.
id: optional string or null The unique ID of the configuration update item.
reasoning: optional { effort } Updates to reasoning configuration. Only effort is supported.
effort: optional ReasoningEffort or null (default: "medium") The reasoning effort to use for subsequent responses until another
configuration update replaces it.
text: optional { format } The text output configuration for subsequent responses.
format: ResponseFormatText or ResponseFormatTextJSONSchemaConfig Use a JSON schema for structured output or text to return to plain text.
One of the following: ResponseFormatText { type } Default response format. Used to generate text responses.
type: "text" The type of response format being defined. Always text.
ResponseFormatTextJSONSchemaConfig { name, schema, type, 2 more } JSON Schema response format. Used to generate structured JSON responses.
Learn more about Structured Outputs.
name: string The name of the response format. Must be a-z, A-Z, 0-9, or contain
underscores and dashes, with a maximum length of 64.
schema: map[unknown] The schema for the response format, described as a JSON Schema object.
Learn how to build JSON schemas here.
type: "json_schema" The type of response format being defined. Always json_schema.
description: optional string A description of what the response format is for, used by the model to
determine how to respond in the format.
strict: optional boolean or null (default: false) Whether to enable strict schema adherence when generating the output.
If set to true, the model will always follow the exact schema defined
in the schema field. Only a subset of JSON Schema is supported when
strict is true. To learn more, read the Structured Outputs
guide.
Reasoning { id, summary, type, 3 more } A description of the chain of thought used by a reasoning model while generating
a response. Be sure to include these items in your input to the Responses API
for subsequent turns of a conversation if you are manually
managing context.
id: string The unique identifier of the reasoning content.
summary: array of SummaryTextContent { text, type } Reasoning summary content.
text: string A summary of the reasoning output from the model so far.
type: "summary_text" (default: "summary_text") The type of the object. Always summary_text.
type: "reasoning" The type of the object. Always reasoning.
content: optional array of { text, type } Reasoning text content.
text: string The reasoning text from the model.
type: "reasoning_text" (default: "reasoning_text") The type of the reasoning text. Always reasoning_text.
encrypted_content: optional string or null The encrypted content of the reasoning item. This is populated by default
for reasoning items returned by POST /v1/responses and WebSocket
response.create requests.
When streaming, use the completed reasoning item and its
encrypted_content from the response.output_item.done event in
subsequent requests. The encrypted_content in
response.output_item.added may be incomplete. This is especially
important when store is false or when using Zero Data Retention.
status: optional "in_progress" or "completed" or "incomplete" The status of the item. One of in_progress, completed, or
incomplete. Populated when items are returned via API.
One of the following: "in_progress" "completed" "incomplete" Compaction { encrypted_content, type, id } A compaction item generated by the v1/responses/compact API.
encrypted_content: string The encrypted content of the compaction summary.
maxLength104857600 type: "compaction" (default: "compaction") The type of the item. Always compaction.
id: optional string or null The ID of the compaction item.
ImageGenerationCall { id, result, status, 9 more } An image generation request made by the model.
id: string The unique ID of the image generation call.
result: string or null The generated image encoded in base64.
status: "in_progress" or "completed" or "generating" or "failed" The status of the image generation call.
One of the following: "in_progress" "completed" "generating" "failed" type: "image_generation_call" The type of the image generation call. Always image_generation_call.
action: optional "generate" or "edit" or "auto" or null The action used for image generation.
One of the following: "generate" "edit" "auto" background: optional "transparent" or "opaque" or "auto" or null The background setting used for generation.
One of the following: "transparent" "opaque" "auto" error: optional { code, message, param, 2 more } Details of a moderation failure, when available.
code: string or null The error code.
message: string A description of the error.
param: string or null The related request parameter.
type: string The error type.
moderation_details: optional { categories, moderation_stage } Public details for a moderation_blocked error, when available.
categories: array of string Public moderation categories. New category values may be added over time.
moderation_stage: "input" or "output" or "unknown" The stage where moderation blocked the image generation call.
One of the following: "input" "output" "unknown" output_format: optional "png" or "webp" or "jpeg" or 2 more or null The output format used for generation.
One of the following: "png" "webp" "jpeg" "gif" "svg" quality: optional "low" or "medium" or "high" or 3 more or null The quality of the image generated by the image generation tool call. One of low, medium, high, xhigh, max, or auto.
One of the following: "low" "medium" "high" "xhigh" "max" "auto" revised_prompt: optional string or null The prompt that was used after any model prompt rewriting.
size: optional string or "1024x1024" or "1024x1536" or "1536x1024" or null The image dimensions as a WIDTHxHEIGHT string, for example 1536x864.
One of the following: string "1024x1024" or "1024x1536" or "1536x1024" The image dimensions as a WIDTHxHEIGHT string, for example 1536x864.
One of the following: "1024x1024" "1024x1536" "1536x1024" svg_usage: optional { input_tokens, model, output_tokens, 2 more } input_tokens: integer model: string output_tokens: integer reasoning_tokens: integer total_tokens: integer CodeInterpreterCall { id, code, container_id, 3 more } A tool call to run code.
id: string The unique ID of the code interpreter tool call.
code: string or null The code to run, or null if not available.
container_id: string The ID of the container used to run the code.
outputs: array of Logs or Image or null The outputs generated by the code interpreter, such as logs or images.
Can be null if no outputs are available.
One of the following: Logs { logs, type } The logs output from the code interpreter.
logs: string The logs output from the code interpreter.
type: "logs" (default: "logs") The type of the output. Always logs.
Image { type, url } The image output from the code interpreter.
type: "image" (default: "image") The type of the output. Always image.
url: string The URL of the image output from the code interpreter.
formaturi status: "in_progress" or "completed" or "incomplete" or 2 more The status of the code interpreter tool call. Valid values are in_progress, completed, incomplete, interpreting, and failed.
One of the following: "in_progress" "completed" "incomplete" "interpreting" "failed" type: "code_interpreter_call" (default: "code_interpreter_call") The type of the code interpreter tool call. Always code_interpreter_call.
LocalShellCall { id, action, call_id, 2 more } A tool call to run a command on the local shell.
id: string The unique ID of the local shell call.
action: { command, env, type, 3 more } Execute a shell command on the server.
command: array of string The command to run.
env: map[string] Environment variables to set for the command.
type: "exec" (default: "exec") The type of the local shell action. Always exec.
timeout_ms: optional integer or null Optional timeout in milliseconds for the command.
user: optional string or null Optional user to run the command as.
working_directory: optional string or null Optional working directory to run the command in.
call_id: string The unique ID of the local shell tool call generated by the model.
status: "in_progress" or "completed" or "incomplete" The status of the local shell call.
One of the following: "in_progress" "completed" "incomplete" type: "local_shell_call" The type of the local shell call. Always local_shell_call.
LocalShellCallOutput { id, output, type, status } The output of a local shell tool call.
id: string The unique ID of the local shell tool call generated by the model.
output: string A JSON string of the output of the local shell tool call.
type: "local_shell_call_output" The type of the local shell tool call output. Always local_shell_call_output.
status: optional "in_progress" or "completed" or "incomplete" or null The status of the item. One of in_progress, completed, or incomplete.
One of the following: "in_progress" "completed" "incomplete" ShellCall { action, call_id, type, 4 more } A tool representing a request to execute one or more shell commands.
action: { commands, max_output_length, timeout_ms } The shell commands and limits that describe how to run the tool call.
commands: array of string Ordered shell commands for the execution environment to run.
max_output_length: optional integer or null Maximum number of UTF-8 characters to capture from combined stdout and stderr output.
timeout_ms: optional integer or null Maximum wall-clock time in milliseconds to allow the shell commands to run.
call_id: string The unique ID of the shell tool call generated by the model.
minLength1 maxLength64 type: "shell_call" (default: "shell_call") The type of the item. Always shell_call.
id: optional string or null The unique ID of the shell tool call. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
environment: optional LocalEnvironment or ContainerReference or null The environment to execute the shell commands in.
One of the following: LocalEnvironment { type, skills } type: "local" (default: "local") Use a local computer environment.
skills: optional array of LocalSkill { description, name, path } An optional list of skills.
maxItems200 description: string The description of the skill.
name: string The name of the skill.
path: string The path to the directory containing the skill.
ContainerReference { container_id, type } container_id: string The ID of the referenced container.
type: "container_reference" (default: "container_reference") References a container created with the /v1/containers endpoint
status: optional "in_progress" or "completed" or "incomplete" or null The status of the shell call. One of in_progress, completed, or incomplete.
One of the following: "in_progress" "completed" "incomplete" ShellCallOutput { call_id, output, type, 4 more } The streamed output items emitted by a shell tool call.
call_id: string The unique ID of the shell tool call generated by the model.
minLength1 maxLength64 output: array of ResponseFunctionShellCallOutputContent { outcome, stderr, stdout } Captured chunks of stdout and stderr output, along with their associated outcomes.
outcome: Timeout or Exit The exit or timeout outcome associated with this shell call.
One of the following: Timeout { type } Indicates that the shell call exceeded its configured time limit.
type: "timeout" (default: "timeout") The outcome type. Always timeout.
Exit { exit_code, type } Indicates that the shell commands finished and returned an exit code.
exit_code: integer The exit code returned by the shell process.
type: "exit" (default: "exit") The outcome type. Always exit.
stderr: string Captured stderr output for the shell call.
maxLength10485760 stdout: string Captured stdout output for the shell call.
maxLength10485760 type: "shell_call_output" (default: "shell_call_output") The type of the item. Always shell_call_output.
id: optional string or null The unique ID of the shell tool call output. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
max_output_length: optional integer or null The maximum number of UTF-8 characters captured for this shell call’s combined output.
status: optional "in_progress" or "completed" or "incomplete" or null The status of the shell call output.
One of the following: "in_progress" "completed" "incomplete" ApplyPatchCall { call_id, operation, status, 3 more } A tool call representing a request to create, delete, or update files using diff patches.
call_id: string The unique ID of the apply patch tool call generated by the model.
minLength1 maxLength64 operation: CreateFile or DeleteFile or UpdateFile The specific create, delete, or update instruction for the apply_patch tool call.
One of the following: CreateFile { diff, path, type } Instruction for creating a new file via the apply_patch tool.
diff: string Unified diff content to apply when creating the file.
maxLength10485760 path: string Path of the file to create relative to the workspace root.
minLength1 type: "create_file" (default: "create_file") The operation type. Always create_file.
DeleteFile { path, type } Instruction for deleting an existing file via the apply_patch tool.
path: string Path of the file to delete relative to the workspace root.
minLength1 type: "delete_file" (default: "delete_file") The operation type. Always delete_file.
UpdateFile { diff, path, type } Instruction for updating an existing file via the apply_patch tool.
diff: string Unified diff content to apply to the existing file.
maxLength10485760 path: string Path of the file to update relative to the workspace root.
minLength1 type: "update_file" (default: "update_file") The operation type. Always update_file.
status: "in_progress" or "completed" The status of the apply patch tool call. One of in_progress or completed.
One of the following: "in_progress" "completed" type: "apply_patch_call" (default: "apply_patch_call") The type of the item. Always apply_patch_call.
id: optional string or null The unique ID of the apply patch tool call. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
ApplyPatchCallOutput { call_id, status, type, 3 more } The streamed output emitted by an apply patch tool call.
call_id: string The unique ID of the apply patch tool call generated by the model.
minLength1 maxLength64 status: "completed" or "failed" The status of the apply patch tool call output. One of completed or failed.
One of the following: "completed" "failed" type: "apply_patch_call_output" (default: "apply_patch_call_output") The type of the item. Always apply_patch_call_output.
id: optional string or null The unique ID of the apply patch tool call output. Populated when this item is returned via API.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
output: optional string or null Optional human-readable log text from the apply patch tool (e.g., patch results or errors).
maxLength10485760 McpListTools { id, server_label, tools, 2 more } A list of tools available on an MCP server.
id: string The unique ID of the list.
server_label: string The label of the MCP server.
tools: array of { input_schema, name, annotations, description } The tools available on the server.
input_schema: unknown The JSON schema describing the tool’s input.
name: string The name of the tool.
annotations: optional unknown or null Additional annotations about the tool.
description: optional string or null The description of the tool.
type: "mcp_list_tools" The type of the item. Always mcp_list_tools.
error: optional string or null Error message if the server could not list tools.
McpApprovalRequest { id, arguments, name, 2 more } A request for human approval of a tool invocation.
id: string The unique ID of the approval request.
arguments: string A JSON string of arguments for the tool.
name: string The name of the tool to run.
server_label: string The label of the MCP server making the request.
type: "mcp_approval_request" The type of the item. Always mcp_approval_request.
McpApprovalResponse { approval_request_id, approve, type, 2 more } A response to an MCP approval request.
approval_request_id: string The ID of the approval request being answered.
approve: boolean Whether the request was approved.
type: "mcp_approval_response" The type of the item. Always mcp_approval_response.
id: optional string or null The unique ID of the approval response
reason: optional string or null Optional reason for the decision.
McpCall { id, arguments, name, 6 more } An invocation of a tool on an MCP server.
id: string The unique ID of the tool call.
arguments: string A JSON string of the arguments passed to the tool.
name: string The name of the tool that was run.
server_label: string The label of the MCP server running the tool.
type: "mcp_call" The type of the item. Always mcp_call.
approval_request_id: optional string or null Unique identifier for the MCP tool call approval request.
Include this value in a subsequent mcp_approval_response input to approve or reject the corresponding tool call.
error: optional McpToolCallError or null The error from the tool call, if any.
output: optional string or null The output from the tool call.
status: optional "in_progress" or "completed" or "incomplete" or 2 more The status of the tool call. One of in_progress, completed, incomplete, calling, or failed.
One of the following: "in_progress" "completed" "incomplete" "calling" "failed" CustomToolCallOutput { call_id, output, type, 2 more } The output of a custom tool call from your code, being sent back to the model.
call_id: string The call ID, used to map this custom tool call output to a custom tool call.
output: string or array of ResponseInputText or ResponseInputImage or ResponseInputFile The output from the custom tool call generated by your code.
Can be a string or an list of output content.
One of the following: StringOutput = string A string of the output of the custom tool call.
OutputContentList = array of ResponseInputText or ResponseInputImage or ResponseInputFile Text, image, or file output of the custom tool call.
One of the following: ResponseInputText { text, type, prompt_cache_breakpoint } A text input to the model.
text: string The text input to the model.
type: "input_text" (default: "input_text") The type of the input item. Always input_text.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputImage { detail, type, file_id, 2 more } An image input to the model. Learn about image inputs.
detail: ImageDetail The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
type: "input_image" (default: "input_image") The type of the input item. Always input_image.
file_id: optional string or null The ID of the file to be sent to the model.
image_url: optional string or null The URL of the image to be sent to the model. A fully qualified URL or base64 encoded image in a data URL.
formaturi prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" ResponseInputFile { type, detail, file_data, 4 more } A file input to the model.
type: "input_file" (default: "input_file") The type of the input item. Always input_file.
detail: optional "auto" or "low" or "high" The detail level of the file to be sent to the model. Use auto to let the system select the detail level; for GPT-5.6 and later models, auto uses high-quality rendering, which may increase input token usage. Use low for lower-cost rendering, or high to render the file at higher quality. Defaults to auto.
One of the following: "auto" "low" "high" file_data: optional string The content of the file to be sent to the model.
file_id: optional string or null The ID of the file to be sent to the model.
file_url: optional string The URL of the file to be sent to the model.
formaturi filename: optional string The name of the file to be sent to the model.
prompt_cache_breakpoint: optional { mode, ttl } Marks the exact end of a reusable prompt prefix. When ttl is omitted, the breakpoint inherits the request’s prompt_cache_options.ttl; the boundary is not rounded to a token block.
mode: "explicit" (default: "explicit") The breakpoint mode. Always explicit.
ttl: optional "30m" or "12h" The minimum lifetime of this cache breakpoint. Defaults to the request TTL.
One of the following: "30m" "12h" type: "custom_tool_call_output" The type of the custom tool call output. Always custom_tool_call_output.
id: optional string The unique ID of the custom tool call output in the OpenAI platform.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") The caller type. Always direct.
Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
minLength1 maxLength64 type: "program" (default: "program") The caller type. Always program.
CustomToolCall { call_id, input, name, 5 more } A call to a custom tool created by the model.
call_id: string An identifier used to map this custom tool call to a tool call output.
input: string The input for the custom tool call generated by the model.
name: string The name of the custom tool being called.
type: "custom_tool_call" The type of the custom tool call. Always custom_tool_call.
id: optional string The unique ID of the custom tool call in the OpenAI platform.
async: optional boolean Whether the custom tool call runs asynchronously.
caller: optional Direct or Program or null The execution context that produced this tool call.
One of the following: Direct { type } type: "direct" (default: "direct") Program { caller_id, type } caller_id: string The call ID of the program item that produced this tool call.
type: "program" (default: "program") namespace: optional string The namespace of the custom tool being called.
CompactionTrigger { type, id } Compacts the current context. Must be the final input item.
type: "compaction_trigger" (default: "compaction_trigger") The type of the item. Always compaction_trigger.
id: optional string or null The unique ID of this compaction trigger.
ItemReference { id, type } An internal identifier for an item to reference.
id: string The ID of the item to reference.
type: optional "item_reference" or null (default: "item_reference") The type of item to reference. Always item_reference.
Program { id, call_id, code, 2 more } id: string The unique ID of this program item.
call_id: string The stable call ID of the program item.
minLength1 maxLength64 code: string The JavaScript source executed by programmatic tool calling.
maxLength10485760 fingerprint: string Opaque program replay fingerprint that must be round-tripped.
maxLength10485760 type: "program" (default: "program") The item type. Always program.
ProgramOutput { id, call_id, result, 2 more } id: string The unique ID of this program output item.
call_id: string The call ID of the program item.
minLength1 maxLength64 result: string The result produced by the program item.
maxLength10485760 status: "completed" or "incomplete" The terminal status of the program output.
One of the following: "completed" "incomplete" type: "program_output" (default: "program_output") The item type. Always program_output.
type: "response.item.create" (default: "response.item.create") The Live client event type. Always response.item.create.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 ResponseCreateEvent { type, event_id } Request a response from the Live session’s Responses backend, or continue a delegated response waiting for tool results. Requires Responses delegation.
type: "response.create" (default: "response.create") The Live client event type. Always response.create.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512 SessionCloseEvent { type, event_id } Request that the Live session close. The terminal session.closed event contains the close reason and final usage.
type: "session.close" (default: "session.close") The Live client event type. Always session.close.
event_id: optional string or null Optional client identifier for correlating this command with a server event’s client_event_id or error.client_event_id.
maxLength512
SidebandServerEvent = TransportRinging { event_id, session_id, type } or TransportAnswered { event_id, session_id, type } or TransportFailed { error, event_id, session_id, type } or 22 moreServer events received by an attached Live sideband WebSocket. Audio deltas are delivered over the primary connection.
SidebandServerEvent = TransportRinging { event_id, session_id, type } or TransportAnswered { event_id, session_id, type } or TransportFailed { error, event_id, session_id, type } or 22 more
Server events received by an attached Live sideband WebSocket. Audio deltas are delivered over the primary connection.
One of the following: TransportRinging { event_id, session_id, type } The outbound SIP provider leg is ringing or providing early media. Delivered only to sideband observers.
event_id: string session_id: string The canonical Live session ID.
type: "transport.ringing" (default: "transport.ringing") TransportAnswered { event_id, session_id, type } The outbound SIP provider leg answered and media is established. Delivered only to sideband observers.
event_id: string session_id: string The canonical Live session ID.
type: "transport.answered" (default: "transport.answered") TransportFailed { error, event_id, session_id, type } An asynchronous outbound SIP setup failure. Delivered only to sideband observers.
error: { code, message, type, param } code: string The call setup failure code.
message: string type: "call_error" (default: "call_error") param: optional string The parameter related to the error, if any. Empty when no parameter applies.
event_id: string session_id: string The canonical Live session ID.
type: "transport.failed" (default: "transport.failed") TransportDtmfReceived { event, event_id, type } A SIP DTMF keypress received from the caller. Delivered only to sideband observers.
event: string pattern^[0-9A-D*#]$ minLength1 maxLength1 event_id: string type: "transport.dtmf.received" (default: "transport.dtmf.received") TransportDtmfSent { event, event_id, type, client_event_id } A SIP DTMF keypress successfully sent to the SIP trunk. Delivered only to sideband observers; this is not a client command.
event: string pattern^[0-9A-D*#]$ minLength1 maxLength1 event_id: string type: "transport.dtmf.sent" (default: "transport.dtmf.sent") client_event_id: optional string The event_id of the client command, when supplied.
SessionStartedEvent { event_id, session, type, client_event_id } Returned when a Live session has started. Contains the resolved session configuration, including server defaults.
event_id: string The unique ID of the Live server event.
session: SessionResource { id, expires_at, model, 9 more } The resolved Live session configuration and server-assigned session metadata.
type: "session.started" (default: "session.started") The event type, always session.started.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
SessionUpdatedEvent { event_id, session, type, client_event_id } Returned when a Live session update is accepted. Contains the resolved session configuration after the update.
event_id: string The unique ID of the Live server event.
session: SessionResource { id, expires_at, model, 9 more } The resolved Live session configuration and server-assigned session metadata.
type: "session.updated" (default: "session.updated") The event type, always session.updated.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InputAudioMutedEvent { event_id, type, client_event_id } Acknowledges a session.input_audio.mute command. An independent input mute sends silence to the model and preserves sideband audio reflection. Fixture playback does not emit this event.
event_id: string The unique ID of the Live server event.
type: "session.input_audio.muted" (default: "session.input_audio.muted") The event type, always session.input_audio.muted.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InputAudioUnmutedEvent { event_id, type, client_event_id } Acknowledges a session.input_audio.unmute command, removing the independent input mute. An active audio fixture still pauses inference until playback finishes. Fixture playback does not emit this event.
event_id: string The unique ID of the Live server event.
type: "session.input_audio.unmuted" (default: "session.input_audio.unmuted") The event type, always session.input_audio.unmuted.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
OutputAudioFixturePlayingEvent { event_id, type, client_event_id } The existing playback path has accepted the validated audio fixture. This acknowledgment does not identify when the listener receives the first sample. The optional client_event_id echoes the play command.
event_id: string The unique ID of the Live server event.
type: "session.output_audio.fixture.playing" (default: "session.output_audio.fixture.playing") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
OutputAudioFixturePlayedEvent { event_id, type, client_event_id } The complete audio fixture has drained from the server mixer and normal inference has resumed. Transport buffers may still contain audio; this event does not prove listener-side completion. The optional client_event_id echoes the play command. This event is not emitted for canceled or failed playback.
event_id: string The unique ID of the Live server event.
type: "session.output_audio.fixture.played" (default: "session.output_audio.fixture.played") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InstructionsAppendedEvent { end_ms, event_id, start_ms, 2 more } Returned when a session.instructions.append command is accepted into the Live session timeline. Acknowledges the appended instructions without guaranteeing that the model has acted on them.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.instructions.appended" (default: "session.instructions.appended") The event type, always session.instructions.appended.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
ThinkingAppendedEvent { end_ms, event_id, start_ms, 2 more } Returned when a session.thinking.append command is accepted into the Live session timeline. Acknowledges the added reasoning context without guaranteeing any spoken output.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.thinking.appended" (default: "session.thinking.appended") The event type, always session.thinking.appended.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
CommentaryAppendedEvent { end_ms, event_id, start_ms, 2 more } Returned when a session.commentary.append command is accepted into the Live session timeline. Acknowledges the added commentary without guaranteeing exact wording or completed audio playback.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.commentary.appended" (default: "session.commentary.appended") The event type, always session.commentary.appended.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InputTranscriptDeltaEvent { delta, end_ms, event_id, 3 more } A transcript fragment for user input audio in the Live session. Accumulate fragments in delivery order; these events do not define complete turns or include a transcript-done event.
delta: string The transcript text fragment for the audio in this time range. Append fragments in delivery order to build the transcript.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.input_transcript.delta" (default: "session.input_transcript.delta") The event type, always session.input_transcript.delta.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
OutputTranscriptDeltaEvent { delta, end_ms, event_id, 3 more } A transcript fragment for assistant output audio in the Live session. Accumulate fragments in delivery order; these events do not define complete turns or include a transcript-done event.
delta: string The transcript text fragment for the audio in this time range. Append fragments in delivery order to build the transcript.
end_ms: integer The end of this event on the Live session timeline, in milliseconds from the beginning of the session. For appended context, this can equal start_ms.
event_id: string The unique ID of the Live server event.
start_ms: integer The start of this event on the Live session timeline, in milliseconds from the beginning of the session.
type: "session.output_transcript.delta" (default: "session.output_transcript.delta") The event type, always session.output_transcript.delta.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
DelegationCreatedEvent { delegation, event_id, offset_ms, 2 more } Returned when the Live model delegates work to your application, Responses, or Decisions. Contains delegation metadata and the position on the session timeline where the work was delegated.
delegation: { id, target, type, response_id } The delegated work identifier and destination. This object contains metadata, not the task text.
id: string The unique ID of the delegation. Use this as delegation_id when replying to client-owned work or correlating backend events.
target: "client" or "responses" or "decisions" Where the Live model delegated the work: client, responses, or decisions.
One of the following: "client" or "responses" or "decisions" Where the Live model delegated the work: client, responses, or decisions.
One of the following: "client" "responses" "decisions" type: "delegation" (default: "delegation") The object type, always delegation.
response_id: optional string The ID of the Responses API response associated with a Responses delegation. Omitted for client delegations.
event_id: string The unique ID of the Live server event.
offset_ms: integer The position on the Live session timeline where the delegation was created, in milliseconds from the beginning of the session.
type: "session.delegation.created" (default: "session.delegation.created") The event type, always session.delegation.created.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
DecisionEvent { delegation_id, event, event_id, 4 more } A Decision API event for a Live delegation or transcript trigger. For a trigger, source_event_id identifies the transcript delta that caused this evaluation.
delegation_id: string or null The Live delegation ID, or null for a transcript trigger.
event: { decision, type } decision: Decision { answers, model, usage } The native Decision API response, including its model, ordered answers, and token usage.
type: "decision.completed" (default: "decision.completed") event_id: string The unique ID of this Live server event.
source_event_id: string or null The event_id of the transcript delta that fired this trigger, or null for a delegation.
source_event_type: "session.input_transcript.delta" or "session.output_transcript.delta" or null One of the following: "session.input_transcript.delta" or "session.output_transcript.delta" One of the following: "session.input_transcript.delta" "session.output_transcript.delta" trigger_id: string or null The Live-assigned ID of this trigger firing, or null for a delegation.
type: "decision.event" (default: "decision.event") DecisionItemCreatedEvent { event_id, item, type, client_event_id } An input item was appended for future decisions.
event_id: string The unique ID of the Live server event.
item: Message or Message or Message or 5 more One of the following: Message { id, content, role, type } id: string content: string or array of InputText or InputImage or InputJSON The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of InputText or InputImage or InputJSON One of the following: InputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" InputJSON { data, type } data: map[unknown] or array of map[unknown] One of the following: LiveDecisionJSONObject = map[unknown] array of map[unknown] type: "input_json" (default: "input_json") role: "user" (default: "user") The message role. Always user.
type: optional "message" (default: "message") Message { id, content, role, type } id: string content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "developer" (default: "developer") type: optional "message" (default: "message") Message { id, content, role, type } id: string content: string or array of DecisionInputText The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") role: "system" (default: "system") type: optional "message" (default: "message") Message { id, content, role, 2 more } id: string content: string or array of OutputText or Refusal The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of OutputText or Refusal One of the following: OutputText { text, type } text: string The text content.
maxLength10485760 type: "output_text" (default: "output_text") Refusal { refusal, type } refusal: string The refusal text.
maxLength10485760 type: "refusal" (default: "refusal") role: "assistant" (default: "assistant") phase: optional "commentary" or "final_answer" or null Labels an assistant message as intermediate commentary (commentary) or the final answer (final_answer). For models like gpt-5.3-codex and beyond, when sending follow-up requests, preserve and resend phase on all assistant messages — dropping it can degrade performance. Not used for user messages.
One of the following: "commentary" "final_answer" type: optional "message" (default: "message") FunctionCall { id, arguments, call_id, 4 more } id: string arguments: string The function arguments as a JSON string.
call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 name: string pattern^[a-zA-Z0-9_.-]+$ minLength1 maxLength128 type: "function_call" (default: "function_call") The item type. Always function_call.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength128 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" FunctionCallOutput { id, call_id, output, 2 more } id: string call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "function_call_output" (default: "function_call_output") The type of the function tool call output. Always function_call_output.
status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCall { id, call_id, input, 4 more } id: string call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 input: string The input to the custom tool call.
name: string The name of the custom tool to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 type: "custom_tool_call" (default: "custom_tool_call") The item type. Always custom_tool_call.
namespace: optional string or null The namespace of the function to call.
pattern^[a-zA-Z0-9_-]+$ minLength1 maxLength64 status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" CustomToolCallOutput { id, call_id, output, 2 more } id: string call_id: string The unique ID of the function tool call generated by the model.
minLength1 maxLength64 output: string or array of DecisionInputText or InputImage The text input to the model.
One of the following: string The text input to the model.
maxLength10485760 array of DecisionInputText or InputImage One of the following: DecisionInputText { text, type } text: string The text input to the model.
maxLength10485760 type: "input_text" (default: "input_text") InputImage { image_url, type, detail } image_url: string A base64-encoded image data URL or a publicly accessible HTTP(S) image URL.
pattern^(data:|https?://) maxLength20971520 type: "input_image" (default: "input_image") detail: optional "low" or "high" or "auto" or "original" or null The detail level of the image to be sent to the model. One of high, low, auto, or original. Defaults to auto.
One of the following: "low" "high" "auto" "original" type: "custom_tool_call_output" (default: "custom_tool_call_output") status: optional "in_progress" or "completed" or "incomplete" or null One of the following: "in_progress" "completed" "incomplete" type: "decision.item.created" (default: "decision.item.created") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
DecisionItemDeletedEvent { event_id, item_id, type, client_event_id } An input item was removed from future decisions.
event_id: string The unique ID of the Live server event.
item_id: string type: "decision.item.deleted" (default: "decision.item.deleted") client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
ResponseEvent { event, event_id, type, 2 more } A streaming Responses API event from a backend delegated to by the Live session. Use the outer delegation_id to associate the nested stream with its Live delegation.
event: map[unknown] The nested Responses streaming event. Dispatch on its type field. Response lifecycle snapshots omit input and clear instructions, tools, and output to keep messages small; consume granular output events for the generated content.
event_id: string The unique ID of the Live server event.
type: "response.event" (default: "response.event") The event type, always response.event.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
delegation_id: optional string or null The Live delegation associated with the nested Responses event. May be null or omitted when the event cannot be correlated with a delegation.
SessionUsageUpdatedEvent { event_id, type, usage, 2 more } Reports cumulative Live audio usage and, when available, the most recent context-window usage. Delegated Responses token usage is reported separately in response.event events.
event_id: string The unique ID of the Live server event.
type: "session.usage.updated" (default: "session.usage.updated") The event type, always session.usage.updated.
usage: SessionUsage { seconds } The cumulative Live audio usage so far.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
context_window: optional { usage_ratio } The latest measured Live context-window usage. Omitted when the context limit is unknown.
usage_ratio: number The latest active context token count divided by the Live model context limit. Can decrease after compaction and may lag between measured audio frames.
SessionClosedEvent { event_id, reason, session, 3 more } Returned after the Live session finishes finalizing, with the close reason, final session snapshot, and cumulative audio usage. A connection closing without this event does not confirm successful finalization.
event_id: string The unique ID of the Live server event.
reason: "close_requested" or "expired" or "content" or 2 more Why the Live session ended: close_requested for an application close or hangup request, expired for the session duration limit, content for a safety filter, remote_hangup for a graceful remote disconnect, or connection_lost for an unexpected primary or upstream disconnection.
One of the following: "close_requested" or "expired" or "content" or 2 more Why the Live session ended: close_requested for an application close or hangup request, expired for the session duration limit, content for a safety filter, remote_hangup for a graceful remote disconnect, or connection_lost for an unexpected primary or upstream disconnection.
One of the following: "close_requested" "expired" "content" "remote_hangup" "connection_lost" session: SessionResource { id, expires_at, model, 9 more } The resolved Live session configuration and server-assigned session metadata.
type: "session.closed" (default: "session.closed") The event type, always session.closed.
usage: SessionUsage { seconds } The final cumulative Live audio usage after session finalization.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
ErrorEvent { error, event_id, type, client_event_id } Reports an error in the Live session, such as an invalid client command. Use error.client_event_id, when present, to identify the command that caused the error.
error: Error { code, message, type, 2 more } Details of the Live error and the client command that caused it, when known.
event_id: string The unique ID of the Live server event.
type: "error" (default: "error") The event type, always error.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.
InfoEvent { code, event_id, message, 2 more } An informational notice about the Live session, such as the event permissions applied to a frontend data channel.
code: string A machine-readable code for the notice, such as data_channel_permissions.
event_id: string The unique ID of the Live server event.
message: string A human-readable explanation of the Live session notice.
type: "info" (default: "info") The event type, always info.
client_event_id: optional string The event_id of the client command associated with this server event, when supplied.