默认情况下，当您向 OpenAI API 发出请求时，我们会先生成模型的完整输出，再通过单个 HTTP 响应返回。生成较长的输出时，等待响应可能需要一些时间。使用流式响应，您可以在模型继续生成完整响应的同时，开始打印或处理已生成的输出。
本指南重点介绍基于服务器发送事件（SSE）的 HTTP 流式传输（stream=true）。如需通过持久 WebSocket 连接传输，并使用 previous_response_id 提供增量输入，请参阅 Responses API 的 WebSocket 模式。
要开始流式传输响应，请在发送到 Responses 端点的请求中设置 stream=True：
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17import { OpenAI } from "openai";
const client = new OpenAI();
const stream = await client.responses.create({
model: "gpt-6-astra",
input: [
{
role: "user",
content: "Say 'double bubble bath' ten times fast.",
},
],
stream: true,
});
for await (const event of stream) {
console.log(event);
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17from openai import OpenAI
client = OpenAI()
stream = client.responses.create(
model="gpt-6-astra",
input=[
{
"role": "user",
"content": "Say 'double bubble bath' ten times fast.",
},
],
stream=True,
)
for event in stream:
print(event)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23package main
import (
"context"
"fmt"
"github.com/openai/openai-go/v3"
"github.com/openai/openai-go/v3/responses"
)
func main() {
client := openai.NewClient()
stream := client.Responses.NewStreaming(context.Background(), responses.ResponseNewParams{
Model: "gpt-6-astra",
Input: responses.ResponseNewParamsInputUnion{OfString: openai.String("Say 'double bubble bath' ten times fast.")},
})
for stream.Next() {
fmt.Println(stream.Current().Type)
}
if err := stream.Err(); err != nil {
panic(err)
}
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.core.http.StreamResponse;
import com.openai.models.responses.ResponseCreateParams;
import com.openai.models.responses.ResponseStreamEvent;
ResponseCreateParams params =
ResponseCreateParams.builder()
.model("gpt-6-astra")
.input("Say 'double bubble bath' ten times fast.")
.build();
try (StreamResponse<ResponseStreamEvent> stream = client.responses().createStreaming(params)) {
stream.stream().forEach(System.out::println);
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18using OpenAI.Responses;
#pragma warning disable OPENAI001
string key = Environment.GetEnvironmentVariable("OPENAI_API_KEY")!;
ResponsesClient client = new(key);
var responses = client.CreateResponseStreamingAsync(
"gpt-6-astra",
"Say 'double bubble bath' ten times fast."
);
await foreach (StreamingResponseUpdate response in responses)
{
if (response is StreamingResponseOutputTextDeltaUpdate delta)
{
Console.Write(delta.Delta);
}
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17require "openai"
openai = OpenAI::Client.new
stream = openai.responses.stream(
model: "gpt-6-astra",
input: [
{
role: "user",
content: "Say 'double bubble bath' ten times fast."
}
]
)
stream.each do |event|
puts(event)
end
Responses API 使用语义事件进行流式传输。每个事件都有预定义模式所规定的类型，因此您可以监听所关注的事件。
如需查看完整的事件类型列表，请参阅流式传输 API 参考。以下是几个示例：
1
2
3
4
5
6
7
8
9for await (const event of stream) {
if (event.type === "response.output_text.delta") {
process.stdout.write(event.delta);
} else if (event.type === "response.completed") {
console.log("\nResponse completed.");
} else if (event.type === "error") {
console.error(event.message);
}
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26StreamingEvent = (
ResponseCreatedEvent
| ResponseInProgressEvent
| ResponseFailedEvent
| ResponseCompletedEvent
| ResponseOutputItemAdded
| ResponseOutputItemDone
| ResponseContentPartAdded
| ResponseContentPartDone
| ResponseOutputTextDelta
| ResponseOutputTextAnnotationAdded
| ResponseTextDone
| ResponseRefusalDelta
| ResponseRefusalDone
| ResponseFunctionCallArgumentsDelta
| ResponseFunctionCallArgumentsDone
| ResponseFileSearchCallInProgress
| ResponseFileSearchCallSearching
| ResponseFileSearchCallCompleted
| ResponseCodeInterpreterInProgress
| ResponseCodeInterpreterCallCodeDelta
| ResponseCodeInterpreterCallCodeDone
| ResponseCodeInterpreterCallInterpreting
| ResponseCodeInterpreterCallCompleted
| Error
)
1type StreamingEvent = responses.ResponseStreamEventUnion
1
2
3
4
5
6
7
8
9
10
11
12import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.core.http.StreamResponse;
import com.openai.models.responses.ResponseCreateParams;
import com.openai.models.responses.ResponseStreamEvent;
ResponseCreateParams params =
ResponseCreateParams.builder().model("gpt-5.5").input("Say hello.").build();
try (StreamResponse<ResponseStreamEvent> stream = client.responses().createStreaming(params)) {
stream.stream().forEach(System.out::println);
}
1
2
3
4
5require "openai"
client = OpenAI::Client.new
stream = client.responses.stream(model: "gpt-5.5", input: "Say hello.")
stream.each { |event| puts(event) }
使用 Chat Completions 进行流式传输相当简单。不过，我们建议使用 Responses API 进行流式传输，因为我们在设计它时就考虑到了流式传输的需求。Responses API 使用语义事件进行流式传输，并且具备类型安全性。
要流式传输补全内容，请在调用 Chat Completions 或旧版 Completions 端点时设置 stream=True。这会返回一个对象，以仅包含数据的服务器发送事件形式流式返回响应。
响应通过事件流分块逐步返回。您可以使用 for 循环遍历事件流，如下所示：
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19import OpenAI from "openai";
const openai = new OpenAI();
const stream = await openai.chat.completions.create({
model: "gpt-6-astra",
messages: [
{
role: "user",
content: "Say 'double bubble bath' ten times fast.",
},
],
stream: true,
});
for await (const chunk of stream) {
console.log(chunk);
console.log(chunk.choices[0].delta);
console.log("****************");
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19from openai import OpenAI
client = OpenAI()
stream = client.chat.completions.create(
model="gpt-6-astra",
messages=[
{
"role": "user",
"content": "Say 'double bubble bath' ten times fast.",
},
],
stream=True,
)
for chunk in stream:
print(chunk)
print(chunk.choices[0].delta)
print("****************")
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24package main
import (
"context"
"fmt"
"github.com/openai/openai-go/v3"
)
func main() {
client := openai.NewClient()
stream := client.Chat.Completions.NewStreaming(context.Background(), openai.ChatCompletionNewParams{
Model: "gpt-6-astra",
Messages: []openai.ChatCompletionMessageParamUnion{
openai.UserMessage("Say 'double bubble bath' ten times fast."),
},
})
for stream.Next() {
fmt.Println(stream.Current())
}
if err := stream.Err(); err != nil {
panic(err)
}
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.core.http.StreamResponse;
import com.openai.models.chat.completions.ChatCompletionChunk;
import com.openai.models.chat.completions.ChatCompletionCreateParams;
ChatCompletionCreateParams params =
ChatCompletionCreateParams.builder()
.model("gpt-6-astra")
.addUserMessage("Say hello.")
.build();
try (StreamResponse<ChatCompletionChunk> stream =
client.chat().completions().createStreaming(params)) {
stream.stream().forEach(System.out::println);
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15using System.ClientModel.Primitives;
using OpenAI.Chat;
string key = Environment.GetEnvironmentVariable("OPENAI_API_KEY")!;
string model = "gpt-6-astra";
ChatClient client = new(model, key);
await foreach (
StreamingChatCompletionUpdate update in client.CompleteChatStreamingAsync(
new UserChatMessage("Say double bubble bath ten times fast.")
)
)
{
Console.WriteLine(ModelReaderWriter.Write(update));
}
1
2
3
4
5
6
7
8
9
10
11
12require "openai"
client = OpenAI::Client.new
stream = client.chat.completions.stream(
model: "gpt-6-astra", messages: [
{
role: :user,
content: "Say hello."
}
]
)
stream.each { |event| puts(event) }
如果您使用我们的 SDK，每个事件都是一个具有明确类型的实例。您也可以使用事件的 type 属性来识别各个事件。
某些关键生命周期事件只会发出一次，而其他事件会在生成响应的过程中多次发出。流式传输文本时，常见的监听事件包括：
- `response.created`
- `response.output_text.delta`
- `response.completed`
- `error`
如需查看可监听事件的完整列表，请参阅流式传输 API 参考。
流式传输聊天补全时，响应包含的是 delta 字段，而不是 message 字段。delta 字段可以包含角色 Token、内容 Token，也可以不包含任何内容。
{ role: 'assistant', content: '', refusal: null }
****************
{ content: 'Why' }
****************
{ content: " don't" }
****************
{ content: ' scientists' }
****************
{ content: ' trust' }
****************
{ content: ' atoms' }
****************
{ content: '?\n\n' }
****************
{ content: 'Because' }
****************
{ content: ' they' }
****************
{ content: ' make' }
****************
{ content: ' up' }
****************
{ content: ' everything' }
****************
{ content: '!' }
****************
{}
****************
如果只想流式传输聊天补全中的文本响应，您可以使用如下代码：
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17import OpenAI from "openai";
const client = new OpenAI();
const stream = await client.chat.completions.create({
model: "gpt-6-astra",
messages: [
{
role: "user",
content: "Say 'double bubble bath' ten times fast.",
},
],
stream: true,
});
for await (const chunk of stream) {
process.stdout.write(chunk.choices[0]?.delta?.content || "");
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18from openai import OpenAI
client = OpenAI()
stream = client.chat.completions.create(
model="gpt-6-astra",
messages=[
{
"role": "user",
"content": "Say 'double bubble bath' ten times fast.",
},
],
stream=True,
)
for chunk in stream:
if chunk.choices[0].delta.content is not None:
print(chunk.choices[0].delta.content, end="")
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26package main
import (
"context"
"fmt"
"github.com/openai/openai-go/v3"
)
func main() {
client := openai.NewClient()
stream := client.Chat.Completions.NewStreaming(context.Background(), openai.ChatCompletionNewParams{
Model: "gpt-6-astra",
Messages: []openai.ChatCompletionMessageParamUnion{
openai.UserMessage("Say 'double bubble bath' ten times fast."),
},
})
for stream.Next() {
if len(stream.Current().Choices) > 0 {
fmt.Print(stream.Current().Choices[0].Delta.Content)
}
}
if err := stream.Err(); err != nil {
panic(err)
}
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.core.http.StreamResponse;
import com.openai.models.chat.completions.ChatCompletionChunk;
import com.openai.models.chat.completions.ChatCompletionCreateParams;
ChatCompletionCreateParams params =
ChatCompletionCreateParams.builder()
.model("gpt-6-astra")
.addUserMessage("Say 'double bubble bath' ten times fast.")
.build();
try (StreamResponse<ChatCompletionChunk> stream =
client.chat().completions().createStreaming(params)) {
stream.stream()
.flatMap(chunk -> chunk.choices().stream())
.flatMap(choice -> choice.delta().content().stream())
.forEach(System.out::print);
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17using OpenAI.Chat;
string key = Environment.GetEnvironmentVariable("OPENAI_API_KEY")!;
string model = "gpt-6-astra";
ChatClient client = new(model, key);
await foreach (
StreamingChatCompletionUpdate update in client.CompleteChatStreamingAsync(
new UserChatMessage("Say double bubble bath ten times fast.")
)
)
{
foreach (ChatMessageContentPart part in update.ContentUpdate)
{
Console.Write(part.Text);
}
}
1
2
3
4
5
6
7
8
9
10
11
12
13require "openai"
client = OpenAI::Client.new
stream = client.chat.completions.stream(
model: "gpt-6-astra",
messages: [
{
role: :user,
content: "Say 'double bubble bath' ten times fast."
}
]
)
stream.text.each { |text| print(text) }
对于流式传输工具调用等更高级的使用场景，请查看以下专题指南：
请注意，在生产应用中流式传输模型输出会增加审核补全内容的难度，因为部分补全内容可能更难评估。这可能会影响获准的使用方式。
如果您在生成请求中请求内容审核分数，分数会在完整输出生成后返回，不会随部分输出的增量一起返回。
