“OpenAPI”

Speech-to-speech voice agent API where the model works on audio directly, with tool calls, over WebRTC or WebSocket.