Overview
- OpenAI rolled out ChatGPT Voice in the desktop app on Thursday, July 23 for paid tiers on macOS and Windows with iOS Remote supported and Android expected later.
- The feature uses GPT‑Live to listen and speak at the same time and can be invoked by a programmable hotkey or a Voice button for hands‑free use.
- On macOS ChatGPT Voice can share an appshot of the frontmost window so the model sees screen context before acting on a user’s request.
- ChatGPT Voice links into ChatGPT Work and Codex so users can start, check, and steer multiple agent tasks that continue running in the background.
- Anthropic released competing voice updates the same day and reporters say GPT‑Live may power a possible OpenAI in‑home speaker, a claim that remains unconfirmed and raises enterprise security and governance questions.