If you talk to ChatGPT out loud, the version you use today is on a clock. ChatGPT Advanced Voice and Standard Voice are both being retired on December 11, 2026, and GPT-Live becomes the only voice mode. Most people will be switched automatically and won’t lose their chats, but there’s one big exception if you use video or screen sharing.
OpenAI announced the change on its Newsroom account on X and in a help center article. Here’s what it means in practice.
Here are the facts worth knowing before the details.
- Retirement date: December 11, 2026 for Advanced Voice and Standard Voice.
- Replacement: GPT-Live, which launched in July and is now the default voice experience.
- Past voice chats stay in your chat history, so there’s nothing to migrate.
- Live video and screen sharing keep running on Advanced Voice for now.
- Custom GPTs lose their older voice options on the same date.
If you only use voice for casual back-and-forth, that last bullet and the video exception are the only places this could trip you up.
What happens to ChatGPT Advanced Voice on December 11
On that date, both older modes disappear from the model picker and from the Voice section of Settings, according to Interesting Engineering. ChatGPT switches affected users over on its own, so you don’t need to start new conversations.
OpenAI says nearly everyone who uses voice has already moved to GPT-Live. It also says voice usage has almost doubled since the July launch. In other words, the company is retiring the modes that most people have stopped touching.
The one group with real homework is anyone who builds custom GPTs. Those lose their legacy voice options on December 11, so check yours before then if voice is part of the experience.
How GPT-Live differs from Standard and Advanced Voice
The three modes handle a conversation in quite different ways, which explains why some people find the switch jarring. This table sums up the differences as OpenAI and the press describe them.
| Mode | How it works | Status |
|---|---|---|
| Standard Voice | Chains three steps: speech to text, a written reply, then text to speech | Retires Dec 11 |
| Advanced Voice | Processes audio directly, but waits for you to finish speaking | Retires Dec 11, except for video and screen sharing |
| GPT-Live | Listens and talks at the same time, so you can interrupt it mid-sentence | Becomes the default voice mode |
OpenAI describes GPT-Live as able to give short acknowledgments while you’re still talking, stay quiet when you seem to be thinking, and pass harder requests to its text models without stopping the conversation. Those are the company’s claims, so how well that works for you is something to test yourself.
Video and screen sharing are the catch
GPT-Live can’t handle visual input yet, so live video and screen sharing stay on Advanced Voice for now. That’s the one reason a version of the old mode survives past December 11.
OpenAI hasn’t given a timeline for moving those features over, according to Interesting Engineering. If you rely on pointing your camera at something or sharing your screen while you talk, nothing changes for you on that date. Just don’t assume that exception is permanent.
Why some users are unhappy, and whether the date could move
The reaction has been loud. Neowin reports user backlash, and Android Headlines says long-time users complain on Reddit and X that GPT-Live jumps in too quickly and cuts off their thoughts. Others say favorite voices like Vale sound different now.
OpenAI’s argument is that running one voice system lets its engineers fix bugs and add features faster. It has also acknowledged that switching models can feel jarring.
There’s precedent for a reversal. TechRadar reported that OpenAI brought Standard Voice back once before after users pushed back hard. Android Headlines notes the December 11 date could change, but that’s speculation. Nothing from OpenAI says it will.
What to do before December 11
You have about a month, so a little preparation goes a long way. These steps cover most situations.
- Spend a few sessions in GPT-Live now, so any dislike of the pacing or voice shows up before the deadline rather than after.
- If you build custom GPTs with voice, test them in the new mode and update anything that depended on the old one.
- If you use video or screen sharing, keep doing so as normal and watch for an OpenAI announcement about those features.
- Keep an eye on your chat history. OpenAI says past conversations remain available and can be continued by text or Live voice.
The goal is simply to avoid a surprise on the day the old options vanish from Settings.
If you like how ChatGPT now reaches out on its own, you may also want our look at ChatGPT Dots texting you on Android and iPhone. And if you’re comparing voice assistants, our piece on Gemini Live offline mode covers where Google is heading.
Frequently Asked Questions
When does ChatGPT Advanced Voice stop working?
December 11, 2026, according to OpenAI. Standard Voice retires on the same day, and GPT-Live takes over as the voice mode.
Will I lose my old voice conversations?
No. Past conversations stay in your chat history, and you can continue them by text or with Live voice. You don’t need to migrate anything.
Can I keep using Advanced Voice for video or screen sharing?
For now, yes. GPT-Live can’t process visual input yet, so those features keep running on Advanced Voice, and OpenAI hasn’t announced when that will change.
Why is OpenAI retiring the older voice modes?
The company says nearly all voice users have already moved to GPT-Live and that maintaining one voice system lets it ship fixes and features faster.
Do I need to do anything before the switch?
Most users don’t, since ChatGPT moves you over automatically. Custom GPT builders should test their voice setups before December 11.
Our take: worth a trial run, not a panic
For most people this is a non-event, because you’ve probably already been moved over. The real risk is for people who loved a particular voice or the slower, patient pacing of Advanced Voice, since that style is going away for good. Try GPT-Live this week, and if it cuts you off too often, send feedback to OpenAI now, because loud complaints have worked before.
Leave a Reply