OpenAI Chat Completions

POST /v1/chat/completions accepts OpenAI-style chat requests. POST /v1/auto/chat/completions forces automatic routing; setting model to auto has the same routing effect on the normal endpoint.

Supported request fields include model, messages, temperature, top_p, max_tokens, max_completion_tokens, stop, stream, tools, tool_choice, response_format, and seed. Unsupported or invalid fields return an OpenAI-compatible error envelope.

from openai import OpenAI

client = OpenAI(base_url="http://127.0.0.1:8742/v1", api_key=divine_token)
completion = client.chat.completions.create(
    model="provider/model",
    messages=[{"role": "user", "content": "Give one word for reliable."}],
    temperature=0,
    max_tokens=16,
)
print(completion.choices[0].message.content)

Streaming

stream = client.chat.completions.create(
    model="provider/model",
    messages=[{"role": "user", "content": "Count to three."}],
    stream=True,
)
for chunk in stream:
    print(chunk.choices[0].delta.content or "", end="")

The HTTP response uses text/event-stream and ends with [DONE]. Once any upstream content has been emitted, Divine Router will not retry or switch providers, preventing duplicated text.

Function tools

completion = client.chat.completions.create(
    model="provider/tool-model",
    messages=[{"role": "user", "content": "Weather in Pune?"}],
    tools=[{
        "type": "function",
        "function": {
            "name": "weather",
            "description": "Get current weather",
            "parameters": {
                "type": "object",
                "properties": {"city": {"type": "string"}},
                "required": ["city"],
            },
        },
    }],
)

Tools, images, and structured-output requirements are capability-validated and never silently removed.