OpenAI has expanded ChatGPT Voice to the desktop app, giving users a more natural way to interact with the assistant through spoken commands. The update lets people direct AI agents and complete computer-based tasks without relying only on typing.
The feature is powered by OpenAI's new voice family, ChatGPT-Live, introduced earlier this month. On desktop, it goes beyond conversation and can support workflows across ChatGPT Work and Codex, including actions that involve multiple steps.
OpenAI says the voice mode can also use computer-use capabilities to browse websites and apps. On macOS, Appshots adds another layer by letting the system read what is on the screen, including alt-text, so the assistant can respond with more context.
The mobile version of ChatGPT Voice focused on smoother, more fluid dialogue, while the desktop release is designed for execution. That means users can issue complex instructions, pause for input when needed, and continue a task in a more interactive flow.
In a demonstration, OpenAI showed a developer using one voice command to create a new thread, open a pull request, and identify the source of a bug. The company also noted that Codex access is available from the iOS app through remote use.
As voice interfaces become more capable, desktop AI may evolve into a more intuitive layer for productivity, coding, and everyday digital work.
