OpenAI WebRTC Audio Session, now with document context

In three linesSimon Willison updated his OpenAI WebRTC tool to support GPT-Realtime-2 model (GPT-5-class reasoning) and added document context feature. Users can now paste text to have audio conversations about specific documents directly in the browser.

Summary generated by Claude — human-verified