Parâmetros da API de Completions de Chat

This document describes the request parameters for the chat completions API, used to generate responses from dialogue models.

Informações da API

• Method: POST
• Path: /v1/chat/completions
• Base URL: https://api.uk72.cn

Autenticação

Authorization: Bearer <APIKey>

Basic Parameters

Parameter	Type	Required	Description
model	string	Sim	Model name, e.g., deepseek-v3
messages	array	Sim	Message list
stream	boolean	Não	Whether to enable streaming response, default false

Message Format

Field	Type	Description
role	string	Role: system, user, assistant, tool
content	string	Message content
name	string	Optional, sender name
tool_calls	array	Tool call list (assistant role)
tool_call_id	string	Tool call ID (tool role)

Generation Control Parameters

Parameter	Type	Default	Description
max_tokens	integer	-	Maximum number of tokens to generate
temperature	number	1.0	Sampling temperature (0~2)
top_p	number	1.0	Nucleus sampling parameter
top_k	integer	-	Top-K sampling
presence_penalty	number	0	Presence penalty (-2~2)
frequency_penalty	number	0	Frequency penalty (-2~2)
repetition_penalty	number	1.0	Repetition penalty

Tool Call Parameters

Parameter	Type	Description
tools	array	Tool definition list
tool_choice	string/object	Tool selection strategy: none, auto, required

Chain of Thought Parameters

Parameter	Type	Description
thinking	boolean	Whether to enable chain of thought
reasoning_effort	string	Reasoning depth control

Request Examples

Basic Conversation

curl -X POST "https://api.uk72.cn/v1/chat/completions" \
  -H "Authorization: Bearer sk-xx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v3",
    "messages": [
      {"role": "user", "content": "Hello"}
    ]
  }'

Streaming Response

curl -X POST "https://api.uk72.cn/v1/chat/completions" \
  -H "Authorization: Bearer sk-xx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v3",
    "messages": [
      {"role": "user", "content": "Hello"}
    ],
    "stream": true
  }'

Parameter Best Practices

• Creative tasks: temperature 0.7-0.9
• Factual answers: temperature 0.1-0.3
• Code generation: temperature 0.1-0.3
• Reduce repetition: repetition_penalty 1.1-1.2
• Conversation scenarios: max_tokens 500-1000

Error Handling

Status Code	Description
400	Parameter validation failed
401	Authentication failed
429	Rate limited
500	Internal server error