🔍 Read the full analysis: OpenAI’s December 11 Voice Changes: What ChatGPT Keeps on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
OpenAI says ChatGPT will remove Advanced and Standard voice from its settings on December 11, 2026, with users switched to GPT-Live. The company says nearly all voice users already use GPT-Live, but some users report differences in interruptions, sound and accessibility. Live video and screen sharing will continue to use Advanced voice until GPT-Live supports them.
OpenAI will retire ChatGPT’s Advanced and Standard voice modes on December 11, 2026, and move users who selected either mode to GPT-Live, the voice system launched in July. The change consolidates ChatGPT’s voice settings, but users have raised concerns about how the newer mode sounds and handles conversation.
OpenAI announced the change on November 10. Its newsroom account on X said voice use has “nearly doubled” since GPT-Live arrived in July and that “virtually all” people using voice in ChatGPT now use GPT-Live. Those figures and the explanation are OpenAI’s; the company did not provide a user count or a breakdown of usage in the cited post.
According to OpenAI’s help-center guidance, as quoted by Interesting Engineering and FourWeekMBA, users can keep using Advanced and Standard voice during the notice period. On December 11, both choices will disappear from Settings > Voice, and people who had selected them will move to Live voice. OpenAI says existing conversations will remain in chat history and can be continued by text or voice; spoken continuations will use GPT-Live.
There are stated exceptions. Live video and screen sharing will continue to use Advanced voice until GPT-Live supports those functions, with no date given for that capability. Voice with custom GPTs will also continue using Advanced voice until the stated December 11 retirement date. The help-center details are reported through those outlets because the pages were not machine-readable in the source account; the retirement announcement and usage explanation were attributed to OpenAI’s X post.
OpenAI’s December 11 Voice Changes What ChatGPT Keeps
Advanced and Standard voice will leave ChatGPT settings, and affected users will move to GPT-Live. Some features will remain on Advanced voice for now.
What changes for users
GPT-Live becomes the general voice option when the two older choices disappear from Settings.
Advanced and Standard remain available during the notice period.
On December 11, both choices are removed and selected users move to GPT-Live.
Existing chats remain in history; spoken continuations use GPT-Live.
How the voice modes work
The modes differ in how they process speech and take turns in conversation.
| Mode | Conversation approach |
|---|---|
| Standard | Speech recognition, language model, then spoken response |
| Advanced | Processes speech directly and waits for the user to stop |
| GPT-Live | Can listen and speak at the same time |
What remains on Advanced
Some capabilities have separate transition conditions.
- Live video and screen sharing stay on Advanced until GPT-Live supports them.
- No timetable for that support was given.
- Voice with custom GPTs continues on Advanced until the retirement date.
- Developers should check API notices separately from the ChatGPT change.
What users are reporting
Some users say GPT-Live interrupts, adds conversational phrases, or sounds different from familiar voices.
What the Voice Switch Changes
For everyday users, the change means one fewer voice choice and a forced move to GPT-Live, rather than a decision to keep using a familiar mode. OpenAI says the shift reflects existing usage, but reports of dissatisfaction indicate that adoption figures do not capture whether people prefer the newer experience or depend on a particular mode.
The distinction may matter to people who use voice for accessibility, as well as to users whose conversations rely on a specific cadence or interaction style. TechRadar reported complaints that GPT-Live interrupts, inserts phrases such as “yeah” and “anyway,” or does not sound like the voices users had been accustomed to. These are user reports, not independently measured findings, and the supplied material does not establish how widespread each complaint is.
The announcement also has a separate implication for developers: the ChatGPT settings change is not itself an API deprecation. OpenAI’s API deprecation page reportedly had no voice entry for November 10. Still, the source material identifies shutdown dates for some older API voice and audio models from January 6 through February 26, 2027. Developers should distinguish those API notices from the ChatGPT product change and check the relevant documentation for their own systems.
Popular right now
Vitamix 5200 Professional-Grade Blender
KitchenAid Classic Series 4.5-Quart Tilt-Head Stand Mixer, Onyx Black
eufy E25 Robot Vacuum and Mop
As an affiliate, we earn on qualifying purchases.
How GPT-Live Differs
The three modes use different approaches to spoken conversation. Standard voice chains together speech recognition, a language model and text-to-speech, according to OpenAI help-center wording quoted by FourWeekMBA. In the description reported by TechRadar, a user’s speech is converted to text, an answer is generated, and that answer is read aloud.
Advanced voice processes speech directly but, according to the same reporting, waits for the user to stop before responding. GPT-Live is described in OpenAI’s developer documentation as a full-duplex voice model that can listen and speak at the same time. OpenAI says the model can pass more demanding reasoning or tool use to a backend agent while continuing the interaction.
OpenAI introduced GPT-Live on July 8 as a new generation of voice models for natural human-AI interaction and said it powers ChatGPT Voice. The company has said it plans to bring features from Advanced and Standard voice into GPT-Live and improve response depth, continuity and personalization, according to reporting by Android Authority and Interesting Engineering. Those are stated plans, not confirmation that every feature or user preference will be matched by December 11.
The retirement notice gives users 31 days between announcement and switch-off. The help-center guidance, as quoted by the outlets, says users do not need to move their old conversations manually. A separate September plan to retire custom GPTs across ChatGPT plans also affects the custom-GPT voice exception; MIXED reported that OpenAI plans to move creators to plugins. For Enterprise workspaces, a migration FAQ reportedly gives December 11 as the date custom GPTs stop running.
Open Questions About GPT-Live
It is not yet clear when GPT-Live will support live video and screen sharing, or whether those functions will work in the same way once they move off Advanced voice. OpenAI has said the existing functions will remain on Advanced until the new model has the capability, but has not supplied a timetable in the material cited.
The scale and specific causes of user dissatisfaction are also uncertain. A forum thread titled “Keep Standard Voice Mode!” had 100 posts when the source account checked on November 12, and TechRadar described opposition as a small but vocal group. Those observations do not provide a representative survey. Some forum participants say they rely on voice for accessibility, but the available material does not quantify how many users are affected or describe individual accommodation needs.
OpenAI’s plans to bring older modes’ features to GPT-Live do not specify which features will arrive by the retirement date. It is also not clear from the cited material whether the reported API shutdown dates apply to every developer using older voice or audio models, or what migration steps each integration may require. The product retirement and API schedules should not be treated as the same announcement.
The December 11 Cutover
Users can continue using Advanced and Standard voice during the announced notice period, which ends on December 11, 2026. On that date, OpenAI says the two choices will be removed from ChatGPT’s voice settings and affected users will be switched to GPT-Live. Their prior chats are expected to remain available in history.
Users who rely on live video, screen sharing or custom-GPT voice will need to watch for further product guidance: the first two features are due to remain on Advanced until GPT-Live supports them, while the custom-GPT exception ends on December 11. Developers using OpenAI audio models should check the API documentation separately for model-specific retirement dates. The next major question is whether OpenAI provides a timetable or details for bringing the remaining Advanced voice capabilities to GPT-Live.
Key Questions
Which ChatGPT voice modes are being retired?
OpenAI says Advanced voice and Standard voice will be removed from ChatGPT’s voice settings on December 11, 2026. GPT-Live will remain as the general voice option.
Will users lose their old voice conversations?
No deletion is described in the reported help-center guidance. Previous conversations are expected to stay in chat history and can be continued by text or voice. A spoken continuation will use GPT-Live.
What happens to live video and screen sharing?
OpenAI’s help-center guidance, as quoted by reporting outlets, says live video and screen sharing will continue to use Advanced voice until GPT-Live supports those capabilities. No timetable for that support was given.
Does this retire OpenAI’s voice API?
Not by itself. The ChatGPT change is a product-settings retirement, not an announced API deprecation. The source material separately reports shutdown dates for some older API voice and audio models between January 6 and February 26, 2027; developers should check OpenAI’s documentation for the models they use.
Why is OpenAI making the change?
OpenAI said voice use has “nearly doubled” since GPT-Live launched in July and that “virtually all” ChatGPT voice users already use it. The company did not provide underlying usage counts in the cited announcement.
Source: ThorstenMeyerAI.com
