This demo requests an ephemeral token directly from OpenAI using the API key you enter below, then creates a WebRTC session and plays remote audio from the realtime model.