ChatGPT Standard and Advanced Voice Are Retiring: What to Change Before December 11

ChatGPT will remove Standard and Advanced voice from its voice model picker on December 11, 2026, switching those users to GPT-Live. Existing conversations stay in history. The important exception is live video and screen sharing, which will continue using Advanced voice until GPT-Live supports them.

OpenAI began notifying users on November 10. Its retirement notice gives you a transition period, not an immediate shutdown. If voice is part of your daily work, use that period to check the tasks you actually depend on rather than assuming the replacement will feel identical.

Can you still select Standard or Advanced voice?

Yes, during the transition. Open Settings > Voice and choose an available mode. On December 11, Standard and Advanced will no longer be selectable there, and accounts using them will default to Live.

This does not erase the conversations you already had. They remain available and can continue in text or voice, although continuing by voice after the change means using GPT-Live.

The history survives; the old model does not. Exporting a conversation saves a record, not a working copy of the voice experience.

Does the retirement remove screen sharing?

OpenAI explicitly says Advanced will keep powering live video and screen sharing until GPT-Live gains those capabilities. Do not interpret the picker retirement as an announcement that these features disappear on December 11. Equally, do not assume Live already supports them.

The current ChatGPT Voice guide describes video and screen sharing for eligible subscribers using Advanced in the iOS and Android apps. Live currently supports neither. OpenAI has not supplied a replacement interface or a GPT-Live video launch date in these pages, so anyone relying on a daily camera workflow should check the in-app notice before switching that workflow permanently.

Test response style and time limits separately

Some users in the OpenAI community discussion describe preferring Standard's longer answers and familiar delivery. Those are individual reports, not a controlled quality comparison. They still identify a useful test: does Live deliver enough substance for the way you work?

Try one familiar task, such as rehearsing a technical interview. Use the same question and context in both modes, then compare whether the answer catches your mistake, explains the reasoning and waits when you need time to think. A faster response is not much help if you have to ask three more questions to get the missing explanation.

You can ask Live to change its pace, tone or response style during a conversation, but preset ChatGPT personalities do not currently apply. For example: “Give me a complete explanation before asking the next question. Keep acknowledgments brief and leave time for me to think.” This is a request to try, not a guaranteed way to reproduce Standard.

For Plus, Live in Chat currently allows three hours over a rolling 24-hour period. It is not a midnight reset. Check your plan's allowance before replacing a long daily workflow; desktop Voice has separate pricing and limits.

Use dictation when you need to check the words first

If the useful part of Standard was speaking a carefully considered question into a text conversation, ChatGPT dictation is worth trying. The microphone records your message and returns a transcription you can edit before sending.

For an interview example, dictate your answer, correct any misheard technical terms, then ask the text conversation to critique it. That creates an explicit checkpoint between speech recognition and reasoning. It is less fluid than a live call and does not recreate Standard Voice, but the checkpoint can be valuable when the exact wording matters.

Save a record without treating export as migration

An export is optional for this voice change because existing chats remain. If you want a separate archive, eligible accounts can request one under Settings > Data controls > Export data. OpenAI's export instructions say delivery can take up to seven days, and the download link expires after 24 hours. Download it using the requesting account and check the ZIP contains the conversations you need.

Self-service export is unavailable for Business, Enterprise and Healthcare workspaces; contact the workspace owner instead. Keep exported files private because they can contain sensitive account information.

The sensible transition test is whether your important task still works, not whether the new voice sounds impressive in a short demo. Compare one familiar conversation now, while you still have a choice, and keep a text-based route for work that needs deliberate review.

I’m Paras. I work at the intersection of AI, engineering and distribution.

I help technical teams turn product changes and tricky workflows into clear explanations their users can actually follow.

Book a 30-minute call.