Supported API subset.
POST /v1/chat/completions accepts a documented subset of the OpenAI request. Anything else is rejected with unsupported_parameter.
Tools, function calling, response_format, logprobs, n greater than 1, images and audio are not supported.
Accepted fields
model (required), messages (required; roles system, user, assistant; string content or text parts), stream, stream_options.include_usage, max_tokens or max_completion_tokens, temperature (0 to 2), top_p (0 to 1], stop (string or up to 4 strings), user.
Defaults and limits
If max_tokens is omitted it defaults to 1,024, capped at the model's published maximum output. Prompts are limited to 400,000 characters and request bodies to 1 MB.
Errors
Errors use the OpenAI envelope {"error": {"message", "type", "code"}}. Stable codes include invalid_api_key, api_key_revoked, api_key_expired, unsupported_parameter, model_not_found, model_unavailable, insufficient_balance, daily_spend_limit_exceeded, key_budget_exceeded and request_failed.