The short version
● Deadline: 11 December 2026 — Advanced Voice and Standard Voice both switch off.
● GPT-Live is the replacement — One mode, no picker. Introduced July 2026.
● You lose the specific voices — The old voice set goes with the old modes.
● Video and screen sharing are unaffected for now — They stay on the older mode until GPT-Live supports them.
Three voice modes become one on 11 December. If you have never thought about which one you were using, nothing will break. If you have, here is exactly what moves.
Side by side
|
Standard Voice |
Advanced Voice |
GPT-Live |
| Status after 11 Dec |
Retired |
Retired |
The only mode |
| Introduced |
Earliest voice mode |
Successor to Standard |
July 2026 |
| What OpenAI says it adds |
— |
— |
Natural conversation with frontier intelligence |
| Voice set |
Its own voices, retiring |
Its own voices, retiring |
GPT-Live voices |
| Video and screen sharing |
Stays on the older mode |
Stays on the older mode |
Not yet supported |
| Action needed |
None |
None |
None — automatic |
OpenAI has not published a date for when video and screen sharing move to GPT-Live.
What actually changes for you
- If you use voice casually — nothing to do. You will notice the voice sounds different and the conversation flows differently.
- If you have a favourite voice — it is part of what retires. The old voice set goes with the old modes, and OpenAI has not said whether equivalents exist in GPT-Live.
- If you use video or screen sharing — unaffected on 11 December, but it stays on a mode OpenAI has now committed to replacing. Treat it as borrowed time rather than stable.
- If you build on it — this is a deadline. Latency, turn-taking and interruption behaviour differ between models, so anything tuned against Advanced Voice needs retesting before 11 December rather than after.
Why OpenAI is consolidating
Its stated reason is adoption: GPT-Live launched in July and voice usage has nearly doubled since. Three modes means three things to maintain and three sets of behaviour to document. One means every improvement lands everywhere.
The cost is the one every consolidation carries. Users who chose a specific mode chose it for a reason, and that reason is not preserved by a replacement that is better on average.
If you are reassessing rather than absorbing
A forced change is a reasonable moment to check the alternatives. Grok Voice, GPT-Live and Gemini Live compared on cost and limits covers what each gives you per minute and where the caps sit.
Frequently asked questions
What is the exact date?
11 December 2026. Advanced Voice and Standard Voice both retire from ChatGPT Voice on that date.
Will I lose access to voice entirely?
No. GPT-Live replaces both and is already available. The transition is automatic.
Can I keep using my current voice?
The voices belonging to Standard and Advanced retire with those modes. OpenAI has not said whether equivalent voices exist in GPT-Live.
What about video and screen sharing?
They remain on the older mode until GPT-Live supports them. No date has been given for that transition.
Should developers do anything before 11 December?
Yes. Retest anything tuned to Advanced Voice behaviour — latency, turn-taking and interruption handling differ between voice models, and after the retirement there is no fallback to compare against.