ChatGPT adds new Live voices, with a screen-sharing tradeoff

You can try a different voice for a spoken creative brief, but check what you need the conversation to do first. OpenAI’s October 15, 2026 rollout brings GPT-Live-1.1 and its smaller Quip model to ChatGPT Voice. It describes more natural accents and adds voices across six languages, plus Bubbie in English. The voices shown vary by account and region. Dated release notes.

A conversation and a screen review need different settings

Live can listen while speaking and work with eligible connected apps. It cannot receive live video or screen sharing. For a phone conversation about what is on your screen, the supported option remains Advanced. Choose the experience under Settings → Voice before starting the review. Voice options and sharing support.

For a proposed first check, read a short project brief containing a client name, a deadline and a change of direction midway through. Ask for the brief back. Does it preserve the name and correction, and can you interrupt comfortably? That tells you more about its fit for your work than a pleasant opening sentence. We have not run this listening comparison.

Check the allowance for the experience you use

The Voice in Chat table measures usage over a rolling 24 hours: Free has limited Quip access, Go lists three hours of Quip, and Plus lists three hours of GPT-Live-1.1. Work tasks also use normal Work allowances; desktop voice has separate terms. Workspace settings, client version and region can affect access. Current usage rules.

Our ChatGPT guide covers the wider writing workflow. For a recorded interview rather than a live conversation, see audio uploads and working notes. Start with the task you need to finish, then choose a voice that helps you finish it.