ChatCompletionsRequest

Request payload for the Chat Completions HTTP API.

This request is 100% compatible with OpenAI's Chat Completions API. All fields follow the official OpenAI specification exactly.

Stateless Operation:

This API is fully stateless, just like OpenAI's standard API. The client is responsible for maintaining conversation history and sending the complete messages array with each request.

Example Request:

`{
"model": "gpt-4",
"messages": [
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "What's the weather in Hangzhou?"},
{"role": "assistant", "content": "Let me check...", "tool_calls": [...]},
{"role": "tool", "tool_call_id": "call_xxx", "content": "25°C sunny"},
{"role": "user", "content": "Thanks! And tomorrow?"}
],
"stream": false
}
`

Tool Call Flow:

When the assistant calls tools, the response includes tool_calls. The client should:

• Append the assistant message (with tool_calls) to history

• Append tool result messages for each tool call

• Send the updated history in the next request