OpenAI Brings Voice Control to ChatGPT Work on Mobile and Web

OpenAI expanded Voice across ChatGPT on September 23, bringing plugin support to spoken conversations and adding Voice to ChatGPT Work on web and mobile. The update lets eligible users start multi-step Work tasks by speaking, while Voice conversations can now call connected apps and display written results alongside the audio.

For Plus and Pro subscribers with access to Work, that means tasks such as drafting documents, creating presentations or spreadsheets, summarizing information from connected apps, and operating a browser can now begin through voice on a phone or the web. If a task continues after the voice session ends, users can pick it up in text.

The update also broadens what Voice can do inside a standard ChatGPT conversation. Live Voice now supports plugins across web, iOS, and Android, allowing users to ask ChatGPT to work with apps already connected to their account. Existing permissions, workspace restrictions, and usage limits still apply.

Some actions still require interaction on screen. When a plugin needs approval, ChatGPT presents controls for the user to review and approve or reject the action. OpenAI says approvals cannot currently be completed by voice alone.

Free and Go users can use Voice with the plugins and connected apps available under their plans, but they do not get the full Work experience. Work remains available on paid plans that include access to the feature.

The rollout builds on GPT-Live-1, which OpenAI introduced for paid ChatGPT users in July, while Free users received GPT-Live-1 mini. OpenAI designed the models to support more natural spoken exchanges, including simultaneous listening and speaking, while also working with features such as web search, memory, and supported visual widgets.

OpenAI followed that release with ChatGPT Work, its agent for longer, multi-step assignments. Work can operate across connected apps and files and produce outputs including documents, spreadsheets, presentations, reports, and Sites. Voice support arrived in Work and Codex on the desktop app later in July, and the September update extends Voice in Work to web and mobile.

Voice sessions are also becoming less isolated from the rest of ChatGPT. Written responses can appear inside the same conversation while users are speaking, and users can switch between text and voice without starting over. Work tasks initiated by voice can likewise continue after the call has ended.

OpenAI lists three Voice options. Live uses GPT-Live-1 or GPT-Live-1 mini depending on the plan. Advanced retains earlier mobile features such as video and screen sharing, while Standard uses a turn-by-turn format that transcribes speech before generating a response.

Live itself still has limitations. OpenAI says it does not currently support video or screen sharing, and it cannot find or attach files directly from a user’s ChatGPT Library. Supported files can still be added manually when the account allows it.

The company also applies existing Voice limits and Work usage rules to the new setup. Tasks started by speaking count against the same Work usage as tasks started by typing, while Live Voice remains subject to plan-specific limits.

The September release makes Voice less of a separate conversational mode and more of an input method for the rest of ChatGPT. Users can now move from a spoken request into plugins, connected apps, or longer Work tasks without first switching to a keyboard, while still retaining on-screen controls for actions that require review.

See more here:

This analysis is based on reporting from Unite.ai.

Image courtesy of OpenAI.

This article was generated with AI assistance and reviewed for accuracy and quality.