ChatGPT will drop Advanced and Standard Voice on December 11

OpenAI will retire Advanced and Standard voice from ChatGPT Voice on December 11, 2026, according to a new entry in its ChatGPT release notes dated November 10. GPT-Live, introduced in July, becomes the only voice model, apart from live video and screen sharing.

The release notes tie the retirement to GPT-Live’s uptake

The entry describes GPT-Live as a model that pairs natural conversation with what OpenAI calls frontier intelligence. According to the company, usage of voice has almost doubled since the July launch, and nearly everyone who talks to ChatGPT now does so through GPT-Live. OpenAI gives that as the reason for the change: it wants a simpler voice experience and a single model to improve.

The one exception is camera and screen features. Live video and screen sharing will stay on Advanced voice until GPT-Live is able to handle them. The notes give no date for that, so Advanced voice survives in a narrow role after December 11, and the page does not say whether it keeps a visible name in the app.

OpenAI also says it is carrying lessons from both older modes into GPT-Live. It names three areas still being worked on: how thorough and detailed answers are, whether a conversation stays coherent across sessions, and personalization. Read plainly, that is an admission that some users may find the older voices better in certain respects. The page does not say which. The notes point to a Help Center article for what users should expect, and the release-note text itself lists no plan or region differences.

The usage figures are OpenAI’s own. The entry gives no absolute numbers, no baseline for the doubling and no share for the small group still on the old modes.

Who is affected and what changes for users

Anyone who still picks Advanced or Standard voice in settings will have to move to GPT-Live after December 11. Since the notes describe almost all voice users as already there, the practical effect is limited to a minority. The notes do not say how that minority is distributed across Free, Plus and Pro plans.

Earlier in October, OpenAI revised the model again with GPT-Live-1.1 and a smaller version called GPT-Live-1.1 Quip, along with new voices for French, German, Hindi, Japanese, Korean and Mexican Spanish, and an English voice named Bubbie, according to the same release notes. So the model that remains has already been updated since launch.

A retirement OpenAI tried once before

This is not the first attempt to remove Standard voice. In 2025, OpenAI planned to retire it in September and fold Advanced voice into a single ChatGPT Voice option. TechRadar reported that users complained about losing it, and later that OpenAI brought it back after the pushback, noting it could still replace the mode in future.

The difference this time is the replacement. GPT-Live launched on July 8, 2026, and outlets covering the launch, including VentureBeat, described it as a full-duplex model that can listen and speak at the same time. Launch coverage also said it hands tasks needing search or deeper reasoning to a stronger model in the background, GPT-5.5 at launch, and that video and screen sharing were not supported then. That last gap is why Advanced voice is not disappearing entirely.

The pattern fits how OpenAI has handled its own interface: it trims legacy options once usage has moved. In September, the same release notes said it was retiring automatic switching from Instant to Thinking for Plus and Pro users. Our earlier coverage of ChatGPT products, such as the Business Hub profile and customer-facing agent in beta, shows the same push to consolidate around newer features.

What the announcement does not settle

The notes do not say whether Standard voice’s distinctive behavior, or voices tied to it, will have an equivalent in GPT-Live. They also do not explain what happens to saved voice conversations or whether custom settings carry over; those answers sit in the Help Center article rather than the release note. Users who rely on voice for long, detailed sessions have a month to test GPT-Live against the modes they use now. If last year’s reversal is a guide, feedback during that window is the one thing OpenAI has previously responded to.