Skip to content

Responses ​

兼容 OpenAI Responses。Codex CLI 使用这个接口。GPT 模型(如 gpt-6.1-sol、gpt-5.5)在 gpt、codex_gpt、az_gpt 分组都可以用。

POSThttps://api.codingrouter.ai/v1/responses

鉴权 ​

Authorization: Bearer <令牌>

请求参数 ​

参数类型必填说明
modelstring是模型名
inputstring / array是一段文本,或消息数组(可包含图片,见下文)
instructionsstring否系统指令
max_output_tokensinteger否最多生成多少 token,包含推理 token
reasoningobject否{"effort": "low" | "medium" | "high"}
streamboolean否true 时以 SSE 事件流返回
toolsarray否可调用的函数,格式见下文
tool_choicestring / object否auto、none、required 或指定函数
temperature、top_pnumber否采样参数,推理模型通常忽略

请求示例 ​

bash
curl https://api.codingrouter.ai/v1/responses \
  -H "Authorization: Bearer $CODINGROUTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-6.1-sol",
    "input": "用三点概括 HTTP/2 相比 HTTP/1.1 的改进"
  }'
python
from openai import OpenAI

client = OpenAI(base_url="https://api.codingrouter.ai/v1", api_key="sk-你的令牌")

resp = client.responses.create(
    model="gpt-6.1-sol",
    input="用三点概括 HTTP/2 相比 HTTP/1.1 的改进",
)
print(resp.output_text)
js
import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.codingrouter.ai/v1",
  apiKey: process.env.CODINGROUTER_KEY,
});

const resp = await client.responses.create({
  model: "gpt-6.1-sol",
  input: "用三点概括 HTTP/2 相比 HTTP/1.1 的改进",
});
console.log(resp.output_text);

响应示例 ​

json
{
  "id": "resp_...",
  "object": "response",
  "status": "completed",
  "model": "gpt-6.1-sol",
  "output": [{
    "type": "message",
    "role": "assistant",
    "content": [{"type": "output_text", "text": "1. **多路复用**:……"}]
  }],
  "usage": {
    "input_tokens": 22,
    "input_tokens_details": {"cached_tokens": 0},
    "output_tokens": 115,
    "output_tokens_details": {"reasoning_tokens": 0},
    "total_tokens": 137
  }
}

SDK 的 output_text 会把所有文本拼好。

推理强度 ​

json
{
  "model": "gpt-6.1-sol",
  "input": "证明根号 2 是无理数",
  "reasoning": {"effort": "high"}
}

推理 token 计入 output_tokens,并在 output_tokens_details.reasoning_tokens 里单独列出。

流式输出 ​

python
stream = client.responses.create(model="gpt-6.1-sol", input="写一段自我介绍", stream=True)
for event in stream:
    if event.type == "response.output_text.delta":
        print(event.delta, end="", flush=True)

事件依次有 response.created、response.output_text.delta(文本增量)、response.output_text.done,最后是带完整响应和用量的 response.completed。

图片输入 ​

json
{
  "model": "gpt-6.1-sol",
  "input": [{
    "role": "user",
    "content": [
      {"type": "input_text", "text": "这张图是什么颜色?"},
      {"type": "input_image", "image_url": "data:image/png;base64,iVBORw0KGgo..."}
    ]
  }]
}

image_url 也可以是公网图片地址。

工具调用 ​

Responses 的函数定义没有外层 function 包装:

json
{
  "model": "gpt-6.1-sol",
  "input": "北京今天天气怎么样?",
  "tools": [{
    "type": "function",
    "name": "get_weather",
    "description": "查询城市天气",
    "parameters": {
      "type": "object",
      "properties": {"city": {"type": "string"}},
      "required": ["city"]
    }
  }]
}

模型调用函数时,output 里会出现 {"type": "function_call", "name": "get_weather", "arguments": "{\"city\":\"北京\"}", "call_id": "call_..."}。