Chat Completions

Chat Completions API 用于多轮对话、文本生成、结构化输出和工具调用前的自然语言处理。接口兼容 OpenAI /v1/chat/completions。

Endpoint

POST http://121.40.161.178/v1/chat/completions

请求示例

curl http://121.40.161.178/v1/chat/completions \
  -H "Authorization: Bearer $PINDUCLOUD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4-flash",
    "messages": [
      {
        "role": "system",
        "content": "你是一个严谨的 API 文档助手。"
      },
      {
        "role": "user",
        "content": "请用三点说明鉴权注意事项。"
      }
    ],
    "temperature": 0.3,
    "max_tokens": 500
  }'

流式输出

curl http://121.40.161.178/v1/chat/completions \
  -H "Authorization: Bearer $PINDUCLOUD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4-flash",
    "messages": [
      {
        "role": "user",
        "content": "输出一个简短的上线检查清单。"
      }
    ],
    "stream": true
  }'

常用参数

参数	类型	说明
model	string	模型名称，例如 deepseek-v4-flash
messages	array	对话消息列表
temperature	number	采样温度，值越高输出越随机
max_tokens	number	最大输出 token 数
stream	boolean	是否启用 SSE 流式响应

响应示例

{
  "id": "chatcmpl_example",
  "object": "chat.completion",
  "created": 1710000000,
  "model": "deepseek-v4-flash",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "1. API Key 仅保存在服务端。\n2. 请求失败时记录错误码和请求 ID。\n3. 对外服务需要限流和审计。"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 32,
    "completion_tokens": 45,
    "total_tokens": 77
  }
}