Skip to main content
Use OpenAI-style content parts for multimodal requests.
The model catalogue marks supported modalities so callers can choose text, image, or audio capable routes.