> ## Documentation Index
> Fetch the complete documentation index at: https://docs.crun.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# OpenAI 兼容语言模型

> 通过 OpenAI 兼容接口接入 CRUN 语言模型，并复用官方 OpenAI SDK。

## 概览

CRUN 在 `https://api.crun.ai/api/v1` 下提供 OpenAI 和 Anthropic 兼容的语言模型网关。
您只需要修改 base URL 和 API Key，就可以继续使用官方 OpenAI SDK，Anthropic SDK 或任何已经兼容 OpenAI 风格 Chat API 的客户端。

<Info>
  当前支持的兼容接口：

  * `GET /models`：列出当前可用的公开模型 ID
  * `POST /responses`：OpenAI最新的接口，推荐用于新的有状态、多模态和工具调用语言模型工作流
  * `POST /chat/completions`：面向现有聊天客户端的 OpenAI 兼容聊天生成接口
  * `POST /messages`：面向 Claude 风格消息客户端的 Anthropic 兼容接口
  * `POST /completions`：旧版文本补全接口；在当前兼容层中不支持流式输出
</Info>

## Base URL

```text theme={null}
https://api.crun.ai/api/v1
```

## 身份验证

您可以使用以下任一方式进行身份验证：

* `Authorization: Bearer YOUR_API_KEY`
* `X-API-KEY: YOUR_API_KEY`

如果使用官方 OpenAI SDK，推荐使用 `Authorization: Bearer YOUR_API_KEY`。

<Warning>
  请妥善保管您的 API Key。切勿在客户端代码或公开代码仓库中暴露它。
</Warning>

## 第 1 步：列出可用模型

模型目录可能会随时间变化。集成或切换模型前，建议始终调用 `/models` 获取最新的公开模型 ID。

<CodeGroup>
  ```bash cURL theme={null} theme={null}
  curl -X GET "https://api.crun.ai/api/v1/models" \
    -H "Authorization: Bearer YOUR_API_KEY"
  ```

  ```python Python theme={null} theme={null}
  import requests

  response = requests.get(
      "https://api.crun.ai/api/v1/models",
      headers={"Authorization": "Bearer YOUR_API_KEY"},
      timeout=30,
  )

  print(response.json())
  ```

  ```javascript Node.js theme={null} theme={null}
  const response = await fetch("https://api.crun.ai/api/v1/models", {
    headers: {
      Authorization: "Bearer YOUR_API_KEY"
    }
  });

  console.log(await response.json());
  ```
</CodeGroup>

### 响应示例

```json theme={null}
{
  "object": "list",
  "data": [
    {
      "id": "gpt-5.4",
      "object": "model",
      "created": 1772294400,
      "owned_by": "OpenAI"
    },
    {
      "id": "claude-sonnet-4-5",
      "object": "model",
      "created": 1772294400,
      "owned_by": "Anthropic"
    },
    {
      "id": "gemini-3.1-pro-preview",
      "object": "model",
      "created": 1772294400,
      "owned_by": "Google"
    }
  ]
}
```

## 第 2 步：发起第一次 Chat Completion 请求

如果您已有 OpenAI Chat Completions 客户端，或需要经典的 `messages` 与 `choices` 响应结构，可以使用 `/chat/completions`。该接口遵循 OpenAI Chat Completions 的请求和响应结构，因此现有 SDK 集成通常只需要替换 base URL。

<CodeGroup>
  ```bash cURL theme={null} theme={null}
  curl -X POST "https://api.crun.ai/api/v1/chat/completions" \
    -H "Authorization: Bearer YOUR_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "gpt-5.4",
      "messages": [
        {
          "role": "system",
          "content": "You are a concise assistant."
        },
        {
          "role": "user",
          "content": "Write a two-line product description for a wireless keyboard."
        }
      ],
      "temperature": 0.7
    }'
  ```

  ```python Python (OpenAI SDK) theme={null} theme={null}
  from openai import OpenAI

  client = OpenAI(
      api_key="YOUR_API_KEY",
      base_url="https://api.crun.ai/api/v1",
  )

  response = client.chat.completions.create(
      model="gpt-5.4",
      messages=[
          {"role": "system", "content": "You are a concise assistant."},
          {"role": "user", "content": "Write a two-line product description for a wireless keyboard."},
      ],
      temperature=0.7,
  )

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

  ```javascript JavaScript (OpenAI SDK) theme={null} theme={null}
  import OpenAI from "openai";

  const client = new OpenAI({
    apiKey: "YOUR_API_KEY",
    baseURL: "https://api.crun.ai/api/v1"
  });

  const response = await client.chat.completions.create({
    model: "gpt-5.4",
    messages: [
      { role: "system", content: "You are a concise assistant." },
      { role: "user", content: "Write a two-line product description for a wireless keyboard." }
    ],
    temperature: 0.7
  });

  console.log(response.choices[0].message.content);
  ```
</CodeGroup>

### 响应示例

```json theme={null}
{
  "id": "chatcmpl_abc123",
  "object": "chat.completion",
  "created": 1772294400,
  "model": "gpt-5.4",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Slim, quiet, and built for all-day comfort.\nA reliable wireless keyboard for focused work at home or on the go."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 24,
    "completion_tokens": 21,
    "total_tokens": 45
  }
}
```

## 第 3 步：使用 Responses API

新的集成建议优先考虑 `/responses`。它支持 OpenAI 更新的 Responses 形态，包括顶层 `instructions`、灵活的 `input`、用于有状态追问的 `previous_response_id`、用于结构化输出的 `text.format`，以及上游模型支持时可用的 `tools`。

<CodeGroup>
  ```bash cURL theme={null} theme={null}
  curl -X POST "https://api.crun.ai/api/v1/responses" \
    -H "Authorization: Bearer YOUR_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "gpt-5.4",
      "instructions": "You are a concise assistant.",
      "input": "Write a two-line product description for a wireless keyboard.",
      "temperature": 0.7
    }'
  ```

  ```python Python (OpenAI SDK) theme={null} theme={null}
  from openai import OpenAI

  client = OpenAI(
      api_key="YOUR_API_KEY",
      base_url="https://api.crun.ai/api/v1",
  )

  response = client.responses.create(
      model="gpt-5.4",
      instructions="You are a concise assistant.",
      input="Write a two-line product description for a wireless keyboard.",
      temperature=0.7,
  )

  print(response.output_text)
  ```

  ```javascript JavaScript (OpenAI SDK) theme={null} theme={null}
  import OpenAI from "openai";

  const client = new OpenAI({
    apiKey: "YOUR_API_KEY",
    baseURL: "https://api.crun.ai/api/v1"
  });

  const response = await client.responses.create({
    model: "gpt-5.4",
    instructions: "You are a concise assistant.",
    input: "Write a two-line product description for a wireless keyboard.",
    temperature: 0.7
  });

  console.log(response.output_text);
  ```
</CodeGroup>

### 响应示例

```json theme={null}
{
  "id": "resp_abc123",
  "object": "response",
  "created_at": 1772294400,
  "status": "completed",
  "model": "gpt-5.4",
  "output": [
    {
      "id": "msg_abc123",
      "type": "message",
      "role": "assistant",
      "content": [
        {
          "type": "output_text",
          "text": "Slim, quiet, and built for all-day comfort.\nA reliable wireless keyboard for focused work at home or on the go."
        }
      ]
    }
  ],
  "output_text": "Slim, quiet, and built for all-day comfort.\nA reliable wireless keyboard for focused work at home or on the go.",
  "usage": {
    "input_tokens": 20,
    "output_tokens": 21,
    "total_tokens": 41
  }
}
```

## 管理对话历史

CRUN 不会为您的应用存储或管理对话状态。
如果希望模型记住之前的对话轮次，您的应用需要自行保存相关消息历史，并在每次新的 `/chat/completions` 请求中重新传入。

<CodeGroup>
  ```bash cURL theme={null} theme={null}
  curl -X POST "https://api.crun.ai/api/v1/chat/completions" \
    -H "Authorization: Bearer YOUR_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "gpt-5.4",
      "messages": [
        {
          "role": "system",
          "content": "You are a concise travel assistant."
        },
        {
          "role": "user",
          "content": "I will spend 2 days in Tokyo. Build a simple itinerary."
        },
        {
          "role": "assistant",
          "content": "Day 1: Asakusa and Ueno. Day 2: Shibuya and Meiji Shrine."
        },
        {
          "role": "user",
          "content": "Update it for rainy weather and add indoor options."
        }
      ],
      "temperature": 0.7
    }'
  ```

  ```python Python (OpenAI SDK) theme={null} theme={null}
  from openai import OpenAI

  client = OpenAI(
      api_key="YOUR_API_KEY",
      base_url="https://api.crun.ai/api/v1",
  )

  conversation_history = [
      {"role": "system", "content": "You are a concise travel assistant."},
      {"role": "user", "content": "I will spend 2 days in Tokyo. Build a simple itinerary."},
      {"role": "assistant", "content": "Day 1: Asakusa and Ueno. Day 2: Shibuya and Meiji Shrine."},
  ]

  conversation_history.append(
      {"role": "user", "content": "Update it for rainy weather and add indoor options."}
  )

  response = client.chat.completions.create(
      model="gpt-5.4",
      messages=conversation_history,
      temperature=0.7,
  )

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

  ```javascript JavaScript (OpenAI SDK) theme={null} theme={null}
  import OpenAI from "openai";

  const client = new OpenAI({
    apiKey: "YOUR_API_KEY",
    baseURL: "https://api.crun.ai/api/v1"
  });

  const conversationHistory = [
    { role: "system", content: "You are a concise travel assistant." },
    { role: "user", content: "I will spend 2 days in Tokyo. Build a simple itinerary." },
    { role: "assistant", content: "Day 1: Asakusa and Ueno. Day 2: Shibuya and Meiji Shrine." },
    { role: "user", content: "Update it for rainy weather and add indoor options." }
  ];

  const response = await client.chat.completions.create({
    model: "gpt-5.4",
    messages: conversationHistory,
    temperature: 0.7
  });

  console.log(response.choices[0].message.content);
  ```
</CodeGroup>

## 流式输出

设置 `stream=true` 后，可以从 `/chat/completions` 接收 Server-Sent Events。
如果未提供 `stream_options.include_usage`，CRUN 会自动启用该选项，以便在流式响应接近结束时包含用量信息。

<CodeGroup>
  ```bash cURL theme={null} theme={null}
  curl -N "https://api.crun.ai/api/v1/chat/completions" \
    -H "Authorization: Bearer YOUR_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "gpt-5.4",
      "stream": true,
      "messages": [
          {"role": "system", "content": "You are a concise assistant."},
          {"role": "user", "content": "Write a two-line product description for a wireless keyboard."}
      ]
    }'
  ```

  ```python Python (OpenAI SDK) theme={null} theme={null}
  from openai import OpenAI

  client = OpenAI(
      api_key="YOUR_API_KEY",
      base_url="https://api.crun.ai/api/v1",
  )

  response = client.chat.completions.create(
      model="gpt-5.4",
      messages=[
          {"role": "system", "content": "You are a concise assistant."},
          {"role": "user", "content": "Write a two-line product description for a wireless keyboard."}
      ],
      stream=True,
      temperature=0.7,
  )

  for chunk in response:
      if chunk.choices:
          print(chunk.choices[0].delta.content, end="")
  ```

  ```javascript JavaScript (OpenAI SDK) theme={null} theme={null}
  import OpenAI from "openai";

  const client = new OpenAI({
    apiKey: "YOUR_API_KEY",
    baseURL: "https://api.crun.ai/api/v1",
  });

  async function main() {
    const stream = await client.chat.completions.create({
      model: "gpt-5.4",
      messages: [
        { role: "system", content: "You are a concise assistant." },
        { role: "user", content: "Write a two-line product description for a wireless keyboard." }
      ],
      temperature: 0.7,
      stream: true,
    });

    for await (const chunk of stream) {
      const content = chunk.choices?.[0]?.delta?.content;
      if (content) {
        process.stdout.write(content);
      }
    }
  }

  main().catch(console.error);
  ```
</CodeGroup>

### 响应示例

```text stream response theme={null} theme={null}
data: {"id":"","object":"chat.completion.chunk","created":0,"model":"gpt-5.4","choices":[{"index":0,"delta":{"role":"assistant","content":"S"}}]}

data: {"id":"","object":"chat.completion.chunk","created":0,"model":"gpt-5.4","choices":[{"index":0,"delta":{"role":"assistant","content":"le"}}]}

data: {"id":"resp_0c628e55cfc393b50069cddfec989c81939cd408ac45285daa","object":"chat.completion.chunk","created":1775099884,"model":"gpt-5.4","choices":[{"index":0,"delta":{"role":"assistant","content":""},"finish_reason":"stop"}],"usage":{"prompt_tokens":17,"completion_tokens":40,"total_tokens":57}}

data: [DONE]
```

## 注意事项

* `max_tokens` 和 `max_completion_tokens` 都受支持。实际值会受所选模型的输出上限约束。
* `/responses` 支持 `max_output_tokens`，实际值同样会受所选模型的输出上限约束。
* 其他 OpenAI 兼容字段也会被接受，并在上游模型支持时透传。
* 计费基于输入和输出 token 用量，并按具体模型设置最低计费值。最低值通常为 0.01。
* 新的有状态或工具调用工作流建议使用 `/responses`。现有聊天客户端仍可继续使用 `/chat/completions`。旧版 `/completions` 接口仅为兼容而提供，不支持 `stream=true`。

## 相关资源

<CardGroup cols={3}>
  <Card title="Responses API" icon="message" href="/zh/models/llm/responses">
    使用灵活输入、有状态追问、结构化输出和工具就绪的 response items 构建应用。
  </Card>

  <Card title="Messages API" icon="message" href="/zh/models/llm/messages">
    使用 Anthropic 兼容 messages、system prompt、工具调用、thinking 和流式事件。
  </Card>

  <Card title="Chat Completions API" icon="message" href="/zh/models/llm/chat-completions">
    查看请求参数、流式输出和 OpenAI 风格错误响应的完整 API 说明。
  </Card>

  <Card title="账户积分" icon="wallet" href="/zh/common-api/get-account-credits">
    在运行大批量语言模型任务前，查询账户剩余积分。
  </Card>

  <Card title="价格" icon="coins" href="https://crun.ai/zh/pricing">
    前往价格页面，比较不同模型的计费规则。
  </Card>
</CardGroup>