OpenAI announced on July 23 that ChatGPT Voice has been added to the desktop app, letting users control their computers by voice and direct multiple AI agents running at the same time. The feature is powered by GPT-Live, OpenAI’s real-time voice model, and began rolling out globally that day on macOS and Windows.
GPT-Live handles conversation and task coordination at once
At the center of ChatGPT Voice is GPT-Live, a real-time voice model that lets ChatGPT listen and respond while coordinating work already in progress inside the app. In practice, users do not need to keep switching between typing and clicking through tasks. Spoken instructions can move the work forward directly.
GPT-Live had previously appeared in a multitasking demo, where it was shown handling flight booking, weather lookups, and itinerary syncing at the same time. This desktop release turns that capability into a single interface for both computer control and agent management.
Voice commands can direct agents in ChatGPT Work and Codex
According to OpenAI, users can operate their computers through voice and manage several AI agents in parallel. Those agents can run inside ChatGPT Work, which is positioned for office collaboration, or Codex, which is used for software development. Users can assign and schedule work through spoken instructions.
That setup allows multiple lines of work to be dispatched in parallel instead of being handled one by one through manual clicks.
Global rollout covers plans from Plus to Enterprise
OpenAI said the desktop version of ChatGPT Voice is available globally starting on the day of the announcement across macOS and Windows. Supported plans include Plus, Pro, Business, Edu, and Enterprise.
The company also posted an approximately 69-second demonstration video with the announcement.