Responses API 流式请求示例

以下示例展示如何使用Response API的 /v1/responses 接口进行流式响应，实现实时内容生成效果。

快速开始

只需要替换 <API-KEY> 为你的实际API密钥即可运行。

curl -X POST "https://api.tokenops.ai/v1/responses" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <API-KEY>" \
  -d '{
    "model": "gpt-5-2025-08-07",
    "input": [
      {
        "role": "user",
        "content": [
          {
            "type": "input_text",
            "text": "请写一首关于春天的诗"
          }
        ]
      }
    ],
    "stream": true,
    "max_output_tokens": 1000
  }' \
  --no-buffer

import requests
import json

# 配置API密钥和基础URL
API_KEY = "<API-KEY>"
BASE_URL = "https://api.tokenops.ai/v1"

def stream_responses_api(user_input):
    url = f"{BASE_URL}/responses"
    headers = {
        "Content-Type": "application/json",
        "Authorization": f"Bearer {API_KEY}"
    }
    
    data = {
        "model": "gpt-5-2025-08-07",
        "input": [
            {
                "role": "user",
                "content": [
                    {
                        "type": "input_text",
                        "text": user_input
                    }
                ]
            }
        ],
        "stream": True,
        "max_output_tokens": 1000
    }
    
    response = requests.post(url, headers=headers, json=data, stream=True)
    
    if response.status_code == 200:
        full_response = ""
        current_event = None
        
        for line in response.iter_lines(decode_unicode=True):
            if not line:
                continue
                
            if line.startswith('event: '):
                current_event = line[7:]  # 去掉 'event: ' 前缀
                
            elif line.startswith('data: '):
                try:
                    data = json.loads(line[6:])  # 去掉 'data: ' 前缀
                    
                    # 处理不同类型的流式事件
                    if current_event == "response.created":
                        print("响应已创建，开始生成内容...")
                        
                    elif current_event == "response.output_text.delta":
                        # 这是最重要的事件，包含实际的文本增量
                        if "delta" in data:
                            text_delta = data["delta"]
                            full_response += text_delta
                            print(text_delta, end="", flush=True)
                    
                    elif current_event == "response.completed":
                        print("\n响应生成完成")
                        break
                        
                except json.JSONDecodeError:
                    continue
                    
        return full_response
    else:
        return f"错误: {response.status_code} - {response.text}"

# 使用示例
if __name__ == "__main__":
    message = "请写一首关于春天的诗"
    print("用户输入:", message)
    print("AI回复: ", end="")
    reply = stream_responses_api(message)
    print(f"\n\n完整回复: {reply}")

const fetch = require('node-fetch');

// 配置API密钥和基础URL
const API_KEY = '<API-KEY>';
const BASE_URL = 'https://api.tokenops.ai/v1';

async function streamResponsesAPI(userInput) {
    const url = `${BASE_URL}/responses`;
    
    const headers = {
        'Content-Type': 'application/json',
        'Authorization': `Bearer ${API_KEY}`
    };
    
    const data = {
        model: 'gpt-5-2025-08-07',
        input: [
            {
                role: 'user',
                content: [
                    {
                        type: 'input_text',
                        text: userInput
                    }
                ]
            }
        ],
        stream: true,
        max_output_tokens: 1000
    };
    
    return new Promise((resolve, reject) => {
        let fullResponse = '';
        let currentEvent = null;
        
        fetch(url, {
            method: 'POST',
            headers: headers,
            body: JSON.stringify(data)
        })
        .then(response => {
            if (!response.ok) {
                throw new Error(`HTTP error! status: ${response.status}`);
            }
            
            const reader = response.body.getReader();
            const decoder = new TextDecoder();
            
            function processStream() {
                return reader.read().then(({ done, value }) => {
                    if (done) {
                        resolve(fullResponse);
                        return;
                    }
                    
                    const chunk = decoder.decode(value);
                    const lines = chunk.split('\n');
                    
                    for (const line of lines) {
                        if (line.startsWith('event: ')) {
                            currentEvent = line.slice(7);
                            
                        } else if (line.startsWith('data: ')) {
                            try {
                                const data = JSON.parse(line.slice(6));
                                
                                if (currentEvent === 'response.created') {
                                    console.log('响应已创建，开始生成内容...');
                                    
                                } else if (currentEvent === 'response.output_text.delta') {
                                    if (data.delta) {
                                        fullResponse += data.delta;
                                        process.stdout.write(data.delta);
                                    }
                                    
                                } else if (currentEvent === 'response.completed') {
                                    console.log('\n响应生成完成');
                                    resolve(fullResponse);
                                    return;
                                }
                            } catch (error) {
                                // 忽略解析错误
                            }
                        }
                    }
                    
                    return processStream();
                });
            }
            
            return processStream();
        })
        .catch(reject);
    });
}

// 使用示例
(async () => {
    try {
        const message = '请写一首关于春天的诗';
        console.log('用户输入:', message);
        process.stdout.write('AI回复: ');
        const reply = await streamResponsesAPI(message);
        console.log(`\n\n完整回复: ${reply}`);
    } catch (error) {
        console.error('错误:', error.message);
    }
})();

package main

import (
    "bufio"
    "bytes"
    "encoding/json"
    "fmt"
    "io"
    "net/http"
    "strings"
)

const (
    APIKey  = "<API-KEY>"
    BaseURL = "https://api.tokenops.ai/v1"
)

type InputContent struct {
    Type string `json:"type"`
    Text string `json:"text"`
}

type InputMessage struct {
    Role    string         `json:"role"`
    Content []InputContent `json:"content"`
}

type ResponsesRequest struct {
    Model           string         `json:"model"`
    Input           []InputMessage `json:"input"`
    Stream          bool           `json:"stream"`
    MaxOutputTokens int            `json:"max_output_tokens"`
}

type DeltaEvent struct {
    Delta string `json:"delta"`
}

func streamResponsesAPI(userInput string) (string, error) {
    url := fmt.Sprintf("%s/responses", BaseURL)
    
    reqData := ResponsesRequest{
        Model: "gpt-5-2025-08-07",
        Input: []InputMessage{
            {
                Role: "user",
                Content: []InputContent{
                    {
                        Type: "input_text",
                        Text: userInput,
                    },
                },
            },
        },
        Stream:          true,
        MaxOutputTokens: 1000,
    }
    
    jsonData, err := json.Marshal(reqData)
    if err != nil {
        return "", err
    }
    
    req, err := http.NewRequest("POST", url, bytes.NewBuffer(jsonData))
    if err != nil {
        return "", err
    }
    
    req.Header.Set("Content-Type", "application/json")
    req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", APIKey))
    
    client := &http.Client{}
    resp, err := client.Do(req)
    if err != nil {
        return "", err
    }
    defer resp.Body.Close()
    
    if resp.StatusCode != 200 {
        body, _ := io.ReadAll(resp.Body)
        return "", fmt.Errorf("API错误: %d - %s", resp.StatusCode, string(body))
    }
    
    scanner := bufio.NewScanner(resp.Body)
    var fullResponse strings.Builder
    var currentEvent string
    
    for scanner.Scan() {
        line := scanner.Text()
        
        if strings.HasPrefix(line, "event: ") {
            currentEvent = strings.TrimPrefix(line, "event: ")
            
        } else if strings.HasPrefix(line, "data: ") {
            dataStr := strings.TrimPrefix(line, "data: ")
            
            if currentEvent == "response.created" {
                fmt.Println("响应已创建，开始生成内容...")
                
            } else if currentEvent == "response.output_text.delta" {
                var deltaEvent DeltaEvent
                if err := json.Unmarshal([]byte(dataStr), &deltaEvent); err == nil {
                    if deltaEvent.Delta != "" {
                        fullResponse.WriteString(deltaEvent.Delta)
                        fmt.Print(deltaEvent.Delta)
                    }
                }
                
            } else if currentEvent == "response.completed" {
                fmt.Println("\n响应生成完成")
                break
            }
        }
    }
    
    return fullResponse.String(), scanner.Err()
}

func main() {
    message := "请写一首关于春天的诗"
    fmt.Println("用户输入:", message)
    fmt.Print("AI回复: ")
    
    reply, err := streamResponsesAPI(message)
    if err != nil {
        fmt.Printf("\n错误: %v\n", err)
        return
    }
    
    fmt.Printf("\n\n完整回复: %s\n", reply)
}

import java.io.BufferedReader;
import java.io.IOException;
import java.io.InputStreamReader;
import java.io.OutputStream;
import java.net.HttpURLConnection;
import java.net.URL;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.fasterxml.jackson.databind.JsonNode;

public class ResponsesStreamingAPI {
    private static final String API_KEY = "<API-KEY>";
    private static final String BASE_URL = "https://api.tokenops.ai/v1";
    
    public static class InputContent {
        public String type;
        public String text;
        
        public InputContent(String type, String text) {
            this.type = type;
            this.text = text;
        }
    }
    
    public static class InputMessage {
        public String role;
        public InputContent[] content;
        
        public InputMessage(String role, InputContent[] content) {
            this.role = role;
            this.content = content;
        }
    }
    
    public static class ResponsesRequest {
        public String model;
        public InputMessage[] input;
        public boolean stream;
        public int max_output_tokens;
        
        public ResponsesRequest(String model, InputMessage[] input, boolean stream, int maxOutputTokens) {
            this.model = model;
            this.input = input;
            this.stream = stream;
            this.max_output_tokens = maxOutputTokens;
        }
    }
    
    public static String streamResponsesAPI(String userInput) throws IOException {
        URL url = new URL(BASE_URL + "/responses");
        HttpURLConnection conn = (HttpURLConnection) url.openConnection();
        
        conn.setRequestMethod("POST");
        conn.setRequestProperty("Content-Type", "application/json");
        conn.setRequestProperty("Authorization", "Bearer " + API_KEY);
        conn.setDoOutput(true);
        
        ObjectMapper mapper = new ObjectMapper();
        ResponsesRequest request = new ResponsesRequest(
            "gpt-5-2025-08-07",
            new InputMessage[]{
                new InputMessage("user", new InputContent[]{
                    new InputContent("input_text", userInput)
                })
            },
            true,
            1000
        );
        
        String jsonBody = mapper.writeValueAsString(request);
        
        try (OutputStream os = conn.getOutputStream()) {
            os.write(jsonBody.getBytes());
        }
        
        if (conn.getResponseCode() == 200) {
            StringBuilder fullResponse = new StringBuilder();
            String currentEvent = null;
            
            try (BufferedReader reader = new BufferedReader(new InputStreamReader(conn.getInputStream()))) {
                String line;
                while ((line = reader.readLine()) != null) {
                    if (line.startsWith("event: ")) {
                        currentEvent = line.substring(7);
                        
                    } else if (line.startsWith("data: ")) {
                        String data = line.substring(6);
                        
                        try {
                            JsonNode event = mapper.readTree(data);
                            
                            if ("response.created".equals(currentEvent)) {
                                System.out.println("响应已创建，开始生成内容...");
                                
                            } else if ("response.output_text.delta".equals(currentEvent)) {
                                if (event.has("delta")) {
                                    String delta = event.get("delta").asText();
                                    fullResponse.append(delta);
                                    System.out.print(delta);
                                }
                                
                            } else if ("response.completed".equals(currentEvent)) {
                                System.out.println("\n响应生成完成");
                                break;
                            }
                        } catch (Exception e) {
                            // 忽略解析错误
                        }
                    }
                }
            }
            
            return fullResponse.toString();
        } else {
            throw new RuntimeException("API错误: " + conn.getResponseCode());
        }
    }
    
    public static void main(String[] args) {
        try {
            String message = "请写一首关于春天的诗";
            System.out.println("用户输入: " + message);
            System.out.print("AI回复: ");
            String reply = streamResponsesAPI(message);
            System.out.println("\n\n完整回复: " + reply);
        } catch (Exception e) {
            System.err.println("错误: " + e.getMessage());
        }
    }
}

<?php

class ResponsesStreamingAPI {
    private $apiKey;
    private $baseUrl;
    
    public function __construct($apiKey, $baseUrl = 'https://api.tokenops.ai/v1') {
        $this->apiKey = $apiKey;
        $this->baseUrl = $baseUrl;
    }
    
    public function streamResponsesAPI($userInput) {
        $url = $this->baseUrl . '/responses';
        
        $data = [
            'model' => 'gpt-5-2025-08-07',
            'input' => [
                [
                    'role' => 'user',
                    'content' => [
                        [
                            'type' => 'input_text',
                            'text' => $userInput
                        ]
                    ]
                ]
            ],
            'stream' => true,
            'max_output_tokens' => 1000
        ];
        
        $postData = json_encode($data);
        
        $context = stream_context_create([
            'http' => [
                'method' => 'POST',
                'header' => [
                    'Content-Type: application/json',
                    'Authorization: Bearer ' . $this->apiKey
                ],
                'content' => $postData
            ]
        ]);
        
        $stream = fopen($url, 'r', false, $context);
        
        if (!$stream) {
            return '无法建立连接';
        }
        
        $fullResponse = '';
        $currentEvent = null;
        
        while (!feof($stream)) {
            $line = trim(fgets($stream));
            
            if (strpos($line, 'event: ') === 0) {
                $currentEvent = substr($line, 7);
                
            } elseif (strpos($line, 'data: ') === 0) {
                $data = substr($line, 6);
                $event = json_decode($data, true);
                
                if ($event) {
                    if ($currentEvent === 'response.created') {
                        echo "响应已创建，开始生成内容...\n";
                        
                    } elseif ($currentEvent === 'response.output_text.delta') {
                        if (isset($event['delta'])) {
                            $delta = $event['delta'];
                            $fullResponse .= $delta;
                            echo $delta;
                            flush();
                        }
                        
                    } elseif ($currentEvent === 'response.completed') {
                        echo "\n响应生成完成\n";
                        break;
                    }
                }
            }
        }
        
        fclose($stream);
        return $fullResponse;
    }
}

// 使用示例
$apiKey = '<API-KEY>';
$api = new ResponsesStreamingAPI($apiKey);

$message = '请写一首关于春天的诗';
echo "用户输入: " . $message . "\n";
echo "AI回复: ";
$reply = $api->streamResponsesAPI($message);
echo "\n\n完整回复: " . $reply . "\n";

?>

流式响应格式

Response API的流式响应使用Server-Sent Events (SSE)格式，每个事件包含两行：event: 和 data:

主要事件类型

1. 响应创建事件

event: response.created
data: {"type":"response.created","sequence_number":0,"response":{"id":"resp_04a2a34a8231e6130068fa23caaaa081939b2e2ce1e99a4492","object":"response","created_at":1761223626,"status":"in_progress"...}}

2. 输出项添加事件

event: response.output_item.added
data: {"type":"response.output_item.added","sequence_number":4,"output_index":1,"item":{"id":"msg_04a2a34a8231e6130068fa23eab57c8193ad2d031223a1220a","type":"message","status":"in_progress","content":[],"role":"assistant"}}

3. 文本增量事件（核心事件）

event: response.output_text.delta
data: {"content_index":0,"delta":"春","item_id":"msg_04a2a34a8231e6130068fa23eab57c8193ad2d031223a1220a","logprobs":[],"output_index":1,"sequence_number":6,"type":"response.output_text.delta"}

event: response.output_text.delta
data: {"content_index":0,"delta":"天","item_id":"msg_04a2a34a8231e6130068fa23eab57c8193ad2d031223a1220a","logprobs":[],"output_index":1,"sequence_number":7,"type":"response.output_text.delta"}

4. 文本输出完成事件

event: response.output_text.done
data: {"content_index":0,"item_id":"msg_04a2a34a8231e6130068fa23eab57c8193ad2d031223a1220a","logprobs":[],"output_index":1,"sequence_number":39,"text":"完整的生成文本","type":"response.output_text.done"}

5. 响应完成事件

event: response.completed
data: {"type":"response.completed","sequence_number":42,"response":{"id":"resp_04a2a34a8231e6130068fa23caaaa081939b2e2ce1e99a4492","object":"response","status":"completed","usage":{"input_tokens":67,"output_tokens":1767,"total_tokens":1834}...}}

重要参数说明

请求参数

model: 使用的模型名称，如 “gpt-5-2025-08-07”
input: 输入消息数组，每个消息包含角色和内容
stream: 设置为 true 启用流式输出
max_output_tokens: 限制输出的最大token数量

关键事件字段

sequence_number: 事件序列号，按时间顺序递增
delta: 文本增量内容（在response.output_text.delta事件中）
item_id: 输出项唯一标识符
output_index: 输出项在响应中的索引位置

SSE格式特点

无[DONE]标记: Response API不使用[DONE]标记，而是通过response.completed事件表示结束
event + data格式: 每个事件都有明确的事件类型和对应的数据
sequence_number: 所有事件都有序列号，确保正确的处理顺序

流式输出的优势

实时反馈: 通过response.output_text.delta事件实时获取文本增量
更好的用户体验: 避免长时间等待，提供打字效果
详细状态: 可以跟踪响应的完整生命周期
结构化信息: 每个事件包含丰富的元数据信息

处理流式数据的注意事项

正确解析SSE格式: 需要同时处理event:和data:行
事件类型识别: 根据不同的事件类型执行相应的处理逻辑
序列号处理: 可以使用sequence_number确保事件顺序
文本累积: 通过response.output_text.delta事件的delta字段累积完整文本
完成检测: 通过response.completed事件确认响应结束

实际应用场景

智能对话: 实现类似ChatGPT的实时对话效果
内容创作: 实时显示文章、诗歌、代码生成过程
思维过程展示: 如果启用reasoning模式，可以实时查看AI的思考过程
多步骤任务: 跟踪复杂任务的执行状态和进度

最佳实践示例

def robust_stream_handler(response):
    """健壮的流式响应处理器"""
    full_response = ""
    current_event = None
    
    for line in response.iter_lines(decode_unicode=True):
        if not line.strip():
            continue
            
        if line.startswith('event: '):
            current_event = line[7:]
            
        elif line.startswith('data: ') and current_event:
            try:
                data = json.loads(line[6:])
                
                # 只处理文本增量事件
                if current_event == "response.output_text.delta":
                    if "delta" in data:
                        delta = data["delta"]
                        full_response += delta
                        print(delta, end='', flush=True)
                        
                # 检测完成
                elif current_event == "response.completed":
                    print("\n响应完成")
                    break
                    
            except json.JSONDecodeError:
                # 跳过无效JSON
                continue
                
    return full_response

通过Response API的流式功能，您可以构建更加交互性和实时性的AI应用，为用户提供卓越的体验。

​Responses API 流式请求示例

​快速开始

​流式响应格式

​主要事件类型

​1. 响应创建事件

​2. 输出项添加事件

​3. 文本增量事件（核心事件）

​4. 文本输出完成事件

​5. 响应完成事件

​重要参数说明

​请求参数

​关键事件字段

​SSE格式特点

​流式输出的优势

​处理流式数据的注意事项

​实际应用场景

​最佳实践示例