模型调用 · 聊天补全 API (0.1.0)

Download OpenAPI specification:

Public HTTP contract derived from the current route signatures and Pydantic request models. Internal /api/private routes are excluded. Dynamic controller results remain an open JSON schema.

Chat Completions

OpenAI-compatible chat completion. stream=true returns an SSE stream; otherwise the upstream completion is returned as JSON. Public calls require large_model:execute permission. Request validation errors use the OpenAI error envelope. Provider 401/403/404 errors normalize to HTTP 502; retry exhaustion can return HTTP 503. Streaming failures appear in SSE error frames.

Authorizations:
BearerAuth
Request Body schema: application/json
required

JSON request accepted by the implemented handler signature. Dictionary fields are validated by the controller.

model
string (Model) ^[=~!@#$&%^*()_+`{}\-\[\];:,.\\?<>\'"|/!¥…·()...
top_p
number (Top P) ( 0 .. 1 ]
Default: 1
top_k
integer (Top K) >= 1
Default: 1
temperature
number (Temperature) [ 0 .. 2 ]
Default: 1
presence_penalty
number (Presence Penalty) [ -2 .. 2 ]
Default: 0
frequency_penalty
number (Frequency Penalty) [ -2 .. 2 ]
Default: 0
max_tokens
integer (Max Tokens) >= 10
Default: 1024
max_completion_tokens
integer (Max Completion Tokens) >= 10
required
Array of objects (Messages)
response_format
object (Response Format)
Default: {}
stream
boolean (Stream)
Default: false
cache
boolean (Cache)
Default: false
Stop (string) or Array of Stop (strings) (Stop)
system
Array of objects (System)
Default: []
Array of objects (Tools)
Tool Choice (string) or Tool Choice (object) (Tool Choice)
model_id
string (Model Id)
Default: ""

Responses

Request samples

Content type
application/json
{
  • "model": "string",
  • "top_p": 1,
  • "top_k": 1,
  • "temperature": 1,
  • "presence_penalty": 0,
  • "frequency_penalty": 0,
  • "max_tokens": 1024,
  • "max_completion_tokens": 10,
  • "messages": [
    ],
  • "response_format": { },
  • "stream": false,
  • "cache": false,
  • "stop": "string",
  • "system": [ ],
  • "tools": [
    ],
  • "tool_choice": "string",
  • "model_id": ""
}

Response samples

Content type
null