# 对话补全 (Chat Completions) (/docs/api/chat)





## 接口定义 [#接口定义]

```http
POST https://token.astrumflow.com/v1/chat/completions
Content-Type: application/json
Authorization: Bearer sk-your-astrumflow-token-key
```

***

## 📥 请求参数 [#-请求参数]

| 参数名               | 类型              |   必填  | 说明                                                                     |
| :---------------- | :-------------- | :---: | :--------------------------------------------------------------------- |
| `model`           | `string`        | **是** | 模型名称，如 `GPT-5.6-sol`、`GPT-5.6-terra`、`GPT-5.6-luna`。                   |
| `messages`        | `array`         | **是** | 对话上下文列表，包含 `role` (`system`, `user`, `assistant`, `tool`) 与 `content`。 |
| `stream`          | `boolean`       |   否   | 是否启用流式输出（SSE）。开启后以数据块持续返回，大幅降低首字等待延迟。                                  |
| `temperature`     | `number`        |   否   | 采样温度，介于 `0.0` 到 `2.0`。数值越低越确定严谨，越高越具创造性。默认 `1.0`。                      |
| `top_p`           | `number`        |   否   | 核采样阈值。通常建议与 `temperature` 仅调整其中之一。                                     |
| `max_tokens`      | `integer`       |   否   | 最大输出 Token 数。                                                          |
| `tools`           | `array`         |   否   | 模型可调用的函数/工具列表（Function Calling）。                                       |
| `tool_choice`     | `string/object` |   否   | 控制模型是否必须调用特定工具。默认 `auto`。                                              |
| `response_format` | `object`        |   否   | 强制结构化输出，如 `{"type": "json_object"}`。                                   |

***

## 💻 请求与调用示例 [#-请求与调用示例]

### 1. 标准非流式请求 [#1-标准非流式请求]

<Tabs items="['cURL', 'Python', 'Node.js']">
  <Tab value="cURL">
    ```bash
    curl https://token.astrumflow.com/v1/chat/completions \
      -H "Content-Type: application/json" \
      -H "Authorization: Bearer sk-your-astrumflow-token-key" \
      -d '{
        "model": "GPT-5.6-terra",
        "messages": [
          {"role": "system", "content": "你是一位资深架构师。"},
          {"role": "user", "content": "解释一下什么是 RESTful API？"}
        ],
        "temperature": 0.7
      }'
    ```
  </Tab>

  <Tab value="Python">
    ```python
    from openai import OpenAI

    client = OpenAI(
        base_url="https://token.astrumflow.com/v1",
        api_key="sk-your-astrumflow-token-key"
    )

    response = client.chat.completions.create(
        model="GPT-5.6-terra",
        messages=[
            {"role": "system", "content": "你是一位资深架构师。"},
            {"role": "user", "content": "解释一下什么是 RESTful API？"}
        ]
    )

    print(response.choices[0].message.content)
    ```
  </Tab>

  <Tab value="Node.js">
    ```javascript
    import OpenAI from 'openai';

    const client = new OpenAI({
      baseURL: 'https://token.astrumflow.com/v1',
      apiKey: 'sk-your-astrumflow-token-key',
    });

    const completion = await client.chat.completions.create({
      model: 'GPT-5.6-terra',
      messages: [
        { role: 'system', content: '你是一位资深架构师。' },
        { role: 'user', content: '解释一下什么是 RESTful API？' },
      ],
    });

    console.log(completion.choices[0].message.content);
    ```
  </Tab>
</Tabs>

#### 标准响应示例 (JSON) [#标准响应示例-json]

```json
{
  "id": "chatcmpl-9Xy78z...",
  "object": "chat.completion",
  "created": 1723650000,
  "model": "GPT-5.6-terra",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "RESTful API 是一种基于 REST（Representational State Transfer）架构风格设计的网络接口..."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 28,
    "completion_tokens": 150,
    "total_tokens": 178
  }
}
```

***

## ⚡ 2. 流式输出 (Streaming Response) [#-2-流式输出-streaming-response]

设置 `"stream": true`，模型将以 Server-Sent Events (SSE) 形式逐步返回数据：

<Tabs items="['Python (流式)', 'Node.js (流式)']">
  <Tab value="Python (流式)">
    ```python
    from openai import OpenAI

    client = OpenAI(
        base_url="https://token.astrumflow.com/v1",
        api_key="sk-your-astrumflow-token-key"
    )

    stream = client.chat.completions.create(
        model="GPT-5.6-sol",
        messages=[
            {"role": "user", "content": "写一段 Python 代码实现高性能快速排序。"}
        ],
        stream=True
    )

    for chunk in stream:
        content = chunk.choices[0].delta.content or ""
        print(content, end="", flush=True)
    ```
  </Tab>

  <Tab value="Node.js (流式)">
    ```javascript
    import OpenAI from 'openai';

    const client = new OpenAI({
      baseURL: 'https://token.astrumflow.com/v1',
      apiKey: 'sk-your-astrumflow-token-key',
    });

    const stream = await client.chat.completions.create({
      model: 'GPT-5.6-sol',
      messages: [{ role: 'user', content: '写一段 Python 代码实现高性能快速排序。' }],
      stream: true,
    });

    for await (const chunk of stream) {
      process.stdout.write(chunk.choices[0]?.delta?.content || '');
    }
    ```
  </Tab>
</Tabs>

***

## 🧠 3. 深度思考模型 (Reasoning Content / Thinking) [#-3-深度思考模型-reasoning-content--thinking]

对于 **GPT-5.6-sol** 等具备深度推理思考链的模型，疾旋Token 在返回结果中支持标准的 `reasoning_content` 思考过程透传。

```python
# 提取 GPT-5.6-sol 深度思考链
response = client.chat.completions.create(
    model="GPT-5.6-sol",
    messages=[{"role": "user", "content": "9.11 和 9.8 哪个数更大？请给出严密数学推导。"}]
)

message = response.choices[0].message

# 思考过程
if hasattr(message, 'reasoning_content') and message.reasoning_content:
    print("【深度思考过程】:")
    print(message.reasoning_content)

# 最终回答
print("\n【最终回答】:")
print(message.content)
```

***

## 👁️ 4. 多模态识图 (Vision) [#️-4-多模态识图-vision]

支持向模型发送图片进行视觉理解与分析：

```python
response = client.chat.completions.create(
    model="GPT-5.6-sol",
    messages=[
        {
            "role": "user",
            "content": [
                {"type": "text", "text": "请详细描述并分析这张架构图。"},
                {
                    "type": "image_url",
                    "image_url": {
                        "url": "https://upload.wikimedia.org/wikipedia/commons/thumb/d/dd/Gfp-wisconsin-madison-the-nature-boardwalk.jpg/2560px-Gfp-wisconsin-madison-the-nature-boardwalk.jpg"
                    }
                }
            ]
        }
    ]
)

print(response.choices[0].message.content)
```