GPT Realtime 2.1 Mini

Cost-efficient variant of OpenAI's current speech-to-speech realtime model. Served via the gateway's /v1/realtime WebSocket endpoint with text and audio input/output and function calling.