Starting October 15, ChatGPT began offering a different set of voices to people who talk to the app out loud. OpenAI started rolling out GPT-Live-1.1, an upgrade to the voice mode that adds 13 new voices and more natural accents. Until now, the app offered nine standard voices, all with English-sounding accents, according to Notebookcheck.
The rollout is gradual, so some accounts will see the change in the coming days while others wait. This is a consumer product story, not a policy or safety event. Several details still rest on secondary reports, and the sections below flag where that matters.
What GPT-Live-1.1 changes in ChatGPT voice mode
Reports based on OpenAI's release notes say the update covers ChatGPT Voice and brings three visible changes: 13 new voices, more natural accents, and less of the "checking..." pause when the voice searches the web. Pasquale Pillitteri's coverage and artrealmai.com both describe the search change. The reports do not say whether the search itself got faster or only the way the handoff is spoken.
The release also introduced GPT-Live-1.1 Quip, a smaller version of the model, according to The Rundown. The reports reviewed did not confirm Quip pricing, so anyone weighing the smaller model should wait for a published rate.
The update follows a late-September change. OpenAI gave voice mode access to apps such as calendars and reminders, but messages are sent only after a manual tap, per Notebookcheck.
Notebookcheck also quoted OpenAI's release notes as saying that "the voices you see may vary by region and account." That single line matters more than it looks. Two people on the same plan in different places may not see the same voice menu.
Which languages and voices are new
The 13 new voices break down by language:
- French: two new voices
- German: two new voices, including Sonna and Miko
- Hindi: two new voices
- Japanese: two new voices
- Korean: two new voices, including Haetsal and Hanul
- Mexican Spanish: two new voices
- English: one new voice, named Bubbie
The nine voices that were already available are Arbor, Breeze, Cove, Ember, Juniper, Maple, Sol, Spruce and Vale. The new set moves voice mode beyond English-sounding options for six languages, which is the most visible change for non-English speakers.
The reports describe the accents as more natural. None of the sources reviewed tested how the voices perform across regional accents or dialects, so that description has not been independently measured.
Who gets the update first and what plan limits apply
Voice access is tied to subscription. The amount of daily voice interaction depends on the plan, from free to Pro, according to Root-Nation. Users who still have access only to the previous voices must wait until the new version is activated on their account.
The reports do not say whether paying subscribers get the new voices before free users. What they describe is one gradual rollout that applies across tiers, with timing set account by account. That makes it hard to predict when a given account will see the update.
How to check whether the new voices are on your account
The reports describe a simple check you can run yourself.
- Update the ChatGPT app to the latest version.
- Open Settings, then Voice.
- Look for the new names in the voice list, including Sonna, Miko, Haetsal, Hanul and Bubbie.
- If they are missing, wait and check again later. The rollout is gradual and the list varies by region and account, so a missing voice does not necessarily point to a fault on your device.
What GPT-Live-1.1 means for developers and the API
The API version is not out yet. OpenAI says version 1.1 is coming soon, but it has not given a date, and no API pricing has been announced. Developers should not plan product changes around it this week.
The most recent API pricing in the reports belongs to the earlier GPT-Live-1 model. According to datanorth.ai, it has cost $0.05 per minute for the voice layer since September 10, with the reasoning model billed separately.
That figure is a baseline, not a forecast. Competitors are pricing voice too. xAI launched Grok Voice 3.0 at $0.08 a minute, as USA 24 reported. The two numbers are not directly comparable because billing scopes may differ, and the GPT-Live-1.1 rate is not public.
Language coverage also shapes who should pay attention. Dutch is not among the named languages, so teams building voice features for customer service or language learning should not assume the new voices cover it.
Open questions and what to watch
Several questions remain open:
- Whether all US accounts receive the new voices, or only some plans and regions.
- When the GPT-Live-1.1 API arrives and what it will cost.
- Whether the faster "checking" messages reflect faster search or reasoning, or only a change in how the handoff sounds.
- What Quip costs and how it compares with the full model.
OpenAI already faces regulatory attention elsewhere. The EU opened its first AI Act probes into OpenAI and Anthropic, as USA 24 reported. None of the coverage of GPT-Live-1.1 links that work to voice features, so there is no basis for reading the probes into this update.
This story relies on secondary coverage of OpenAI's release notes. Readers who depend on exact wording or pricing should check OpenAI's own notes before making a development or purchasing decision.
The clearest next signal is the API. If version 1.1 reaches developers with the same voices and accents, it will show whether the upgrade stays a consumer feature or becomes part of OpenAI's developer platform. Until then, the change is a gradual one, and the voices you see depend on where you live and which plan you pay for.
