OpenAI 兼容 API

OpenAI 兼容 API

HeteroFlow V2 推理网关完全兼容 OpenAI API 格式，可直接使用任何 OpenAI SDK 或兼容客户端调用推理服务。

基础信息

• Base URL：http://<server-ip>:3333/v1
• 认证方式：API Key（Header: Authorization: Bearer sk-xxx）
• API Key 来源：创建推理服务时自动生成

Chat Completions

POST /v1/chat/completions

请求示例：

curl http://localhost:3333/v1/chat/completions \
  -H "Authorization: Bearer sk-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d ''{
    "model": "my-service-name",
    "messages": [
      {"role": "system", "content": "你是一个有帮助的AI助手。"},
      {"role": "user", "content": "你好，请介绍一下你自己"}
    ],
    "temperature": 0.7,
    "max_tokens": 512
  }''

Python SDK 调用：

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:3333/v1",
    api_key="sk-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
)

response = client.chat.completions.create(
    model="my-service-name",
    messages=[
        {"role": "user", "content": "你好"}
    ]
)
print(response.choices[0].message.content)

流式响应：

curl http://localhost:3333/v1/chat/completions \
  -H "Authorization: Bearer sk-xxx" \
  -H "Content-Type: application/json" \
  -d ''{"model": "my-service", "messages": [...], "stream": true}''

Models

GET /v1/models

返回当前可用的推理服务列表。

响应示例：

{
  "object": "list",
  "data": [
    {
      "id": "my-service-name",
      "object": "model",
      "owned_by": "heteroflow"
    }
  ]
}

认证

所有请求需要携带 API Key：

Authorization: Bearer sk-xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx

API Key 在创建推理服务时自动生成，可在推理服务列表卡片上查看。

支持的参数

错误响应