mstar.api_server.openai.serving_chat

/v1/chat/completions handler (streaming + non-streaming).

Translates an OpenAI chat request into a submit_request via the model adapter, then maps the resulting modality chunks back to OpenAI shapes: text into message.content, audio into message.audio (base64 WAV), images into image_url data-URL content parts.

Functions

create_chat_completion(api, model_name, ...)

async mstar.api_server.openai.serving_chat.create_chat_completion(api, model_name, adapter, req)