← Docs/

§ API 参考

POST /v1/chat/completions

OpenAI Chat Completions API 兼容. 同时支持 /v1/responses (新版 Responses API). 客户端把 base_url 换成 HiPMM, key 换成 hipmm_sk_, 其它零改.

01

端点

路径
/v1/chat/completions · /v1/responses
Method
POST
Auth
Authorization: Bearer hipmm_sk_...
Stream
✅ SSE
Cost
10 units (managed) / 4 units (BYOM)
Plan
Starter / Pro / Enterprise
02

/v1/chat/completions 请求

application/json

{
  "model": "gpt-4o",
  "messages": [
    { "role": "system",    "content": "You are PMM-8+ aware." },
    { "role": "user",      "content": "Hi" }
  ],
  "stream": true,
  "temperature": 0.7,
  "max_tokens": 1024
}
字段类型必填说明
modelstring✅OpenAI 模型名 (managed: gpt-4o / gpt-4o-mini ...)
messagesarray✅role ∈ {system, user, assistant, tool}
streamboolean—true 走 SSE, 与 OpenAI 100% 一致
temperaturenumber—0-2
max_tokensinteger—输出上限
tools / tool_choice——function calling
response_formatobject—JSON mode / json_schema
03

响应 (非流式)

application/json

{
  "id": "chatcmpl-...",
  "object": "chat.completion",
  "created": 1736899200,
  "model": "gpt-4o",
  "choices": [
    {
      "index": 0,
      "message": { "role": "assistant", "content": "..." },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens":     142,
    "completion_tokens": 568,
    "total_tokens":      710
  }
}
04

流式 (SSE, 与 OpenAI 完全一致)

text/event-stream · 节选

data: {"id":"chatcmpl-...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"role":"assistant","content":""}}]}

data: {"id":"chatcmpl-...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"content":"深"}}]}

data: {"id":"chatcmpl-...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"content":"思"}}]}

...

data: {"id":"chatcmpl-...","choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}

data: [DONE]
05

/v1/responses (新版 Responses API)

OpenAI 2024 推出的 Responses API. HiPMM 同样兼容. 请求体与 chat.completions 略不同 (用 input 替代 messages):

application/json

{
  "model": "gpt-4o",
  "input": [
    { "role": "user", "content": "Hi" }
  ],
  "instructions": "You are PMM-8+ aware.",
  "stream": true
}

响应是 response.created → response.output_item.added → response.output_text.delta → ... → response.completed 事件流, 与 OpenAI 上游一致.

06

客户端示例

python · openai SDK

from openai import OpenAI

client = OpenAI(
    api_key="hipmm_sk_xxx",
    base_url="https://engine.hipmm.dev/v1",  # 唯一改动
)

resp = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hi"}],
)
print(resp.choices[0].message.content)

typescript · openai SDK

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: "hipmm_sk_xxx",
  baseURL: "https://engine.hipmm.dev/v1",
});

const resp = await client.chat.completions.create({
  model: "gpt-4o",
  messages: [{ role: "user", content: "Hi" }],
});
07

BYOM 模式

header 加 X-HiPMM-LLM-Config

Authorization: Bearer hipmm_sk_xxx
X-HiPMM-LLM-Config: {"mode":"byom_proxy","provider":"openai","api_key":"sk-..."}
Content-Type: application/json

提示

provider 可选 openai / deepseek / 任何 OpenAI-compatible. DeepSeek 用 {"provider":"deepseek","base_url":"https://api.deepseek.com/v1","model":"deepseek-v4-pro"}. 见 BYOM 文档.
08

错误码

错误格式与 OpenAI 一致 ({"error":{"message":...,"type":...,"code":...}}), 完整码表见 错误码.