Quick answer: GPT-Live-1.1 is rolling out in ChatGPT Voice with more natural accents, more varied spoken cues while it searches, and new Live voices for French, German, Hindi, Japanese, Korean, Mexican Spanish, Brazilian Portuguese, and English. To check your account, open Settings → Voice, choose Live, then open Voice to select an available voice. Rollout varies by account and region, so a missing voice does not necessarily mean your app is broken. [1] [2]
Two-minute setup checklist
- Update the ChatGPT app, or sign in at ChatGPT.com.
- Open Settings → Voice and select Live.
- Open Voice and choose one of the voices shown for your account.
- Open a new chat, tap the Voice icon, and allow microphone access if prompted.
- Set your main spoken language under Settings → Voice → Language.
- Run the short pronunciation and search tests below before relying on Voice for work.
- Review Settings → Data Controls before discussing sensitive information.
What changed in GPT-Live-1.1?
| Change | What it means |
|---|---|
| GPT-Live-1.1 | The current full Live model for eligible paid plans and workspaces. |
| GPT-Live-1.1 Quip | A smaller version used for limited Free access and the Go plan. |
| New voices | Additional regional-language options plus Bubbie in English; availability can vary. |
| More natural accents | OpenAI says the rollout improves accents across supported languages. |
| Search cues | Live uses a wider range of short phrases to indicate that it is searching. |
| API status | OpenAI’s October 15 announcement said GPT-Live-1.1 API access was coming soon; do not treat the ChatGPT rollout as proof that the 1.1 API model is already generally available. |
OpenAI announced the ChatGPT rollout on October 15, 2026. Its release notes identify GPT-Live-1.1 and GPT-Live-1.1 Quip, while the official Voice guide documents current modes, plan limits, voices, privacy controls, and troubleshooting. OpenAI’s social announcement says the update also applies to Voice in Codex and dots and that API access is coming later. [1] [2] [4]
GPT-Live-1.1 voice list
The following Live voices are listed in OpenAI’s current Voice help page. The page groups Viola and Rio under “Brazil”; this guide describes that group as Brazilian Portuguese rather than implying that “Brazil” is a language. Voices displayed in your account may differ by region and rollout status. [2]
| Language or regional group | New Live voices |
|---|---|
| Brazilian Portuguese | Viola, Rio |
| French | Aube, Rivage |
| German | Sonna, Miko |
| Hindi | Leher, Neer |
| Japanese | Himari (ひまり), Rindō (りんどう) |
| Korean | Haetsal (햇살), Hanul (한울) |
| Mexican Spanish | Tinta, Arrullo |
| English | Bubbie |
OpenAI also continues to list the existing voices Arbor, Breeze, Cove, Ember, Juniper, Maple, Sol, Spruce, and Vale. Changing the voice during a call starts a new Voice call inside the same chat. [2]
How to use GPT-Live-1.1 on mobile and web
On iPhone or Android
- Open ChatGPT and select your profile or settings menu.
- Go to Voice and choose Live if your account shows multiple Voice modes.
- Select Voice, preview the available options, and save your choice.
- Return to a chat and tap the Voice icon in the message bar.
- Allow microphone access when your phone asks.
- Speak normally. Use the microphone control to mute and the exit control to end the call.
On the web
- Sign in at ChatGPT.com using a supported browser.
- Open Settings → Voice, select Live, and choose a voice.
- Open a chat and select the Voice icon in the prompt window.
- Allow browser microphone access, then begin speaking.
Live can combine speech with typed text and supported images in the same chat. It can also use web search, memory, plugins, and connected apps when those features are available to the account. Actions that require approval must be reviewed with on-screen controls; spoken approval is not supported. [2]
Live vs Advanced vs Standard Voice
| Mode | Choose it when | Main limitation |
|---|---|---|
| Live | You want natural interruption, simultaneous listening and speaking, search, memory, text, images, or connected apps. | It does not currently support video or screen sharing. |
| Advanced | You need eligible mobile video or screen sharing. | It is the previous real-time Voice experience rather than GPT-Live-1.1. |
| Standard | You prefer turn-by-turn speech that is transcribed before a response. | Conversation is less fluid than Live. |
The important trap is assuming that selecting a new voice also selects Live. Check the mode and the voice separately. If screen sharing is essential, switch to Advanced on an eligible mobile account instead of repeatedly searching for the control in Live. [2]
Current ChatGPT Voice limits by plan
OpenAI says Voice in Chat is measured on a rolling 24-hour period and displays a notice when the limit is reached. The current help page lists the following allowances; pricing and limits can change, so verify the live plan page before making a purchase decision. [2]
| Plan | Listed Voice in Chat allowance |
|---|---|
| Free | Limited GPT-Live-1.1 Quip access; limit may change |
| Go | 3 hours of GPT-Live-1.1 Quip |
| Plus | 3 hours of GPT-Live-1.1 |
| Pro ($100/month) | 15 hours of GPT-Live-1.1 |
| Pro ($200/month) | Unlimited GPT-Live-1.1, subject to abuse guardrails |
| Business Standard | 3 hours, then 1.25 credits per minute |
| Business Premium | 15 hours, then 1.25 credits per minute |
| K-12 Teachers | 3 hours |
| Enterprise, Edu, and Clinicians on credit pricing | 1.25 credits per minute |
| Enterprise on usage-based USD pricing | $0.05 per minute |
| Legacy Enterprise and Edu | 3 hours |
Voice in Work follows Voice limits and billing while its tasks also count toward normal Work usage. Desktop Voice has separate availability, pricing, and limits. Workspace settings can disable or restrict features even when a plan normally supports them. [2]
A five-minute language and search test
Use this test before a lesson, meeting, accessibility workflow, or customer-facing task:
- Language lock: set the preferred language in Voice settings, then say, “For this conversation, answer only in [language].”
- Name test: read five names, brands, or technical terms you actually use and ask Live to repeat them.
- Number test: dictate a date, phone number, currency amount, and decimal, then check the text in the chat.
- Interruption test: interrupt a long answer with “Stop—give me the two key points.” Confirm that it changes course.
- Search test: ask for a current fact and require the exact source link and publication date. Inspect the source rather than trusting the spoken answer.
- Noise test: repeat one prompt with and without headphones. If interruptions persist, move to a quieter room or use Voice Isolation on a supported iPhone.
Voice transcripts are not guaranteed to be verbatim, and current or location-sensitive answers can be wrong. For a consequential number, date, instruction, or claim, review the written response and its primary source. [2]
How to troubleshoot missing voices or Live mode
- Confirm rollout status: new voices can vary by region and account.
- Update the client: install the latest ChatGPT app or refresh ChatGPT.com.
- Check the mode: go to Settings → Voice and explicitly select Live rather than Standard or Advanced.
- Check plan and workspace controls: plan, region, parental controls, app version, and managed-workspace settings can change what appears.
- Check microphone permission: review app or browser microphone access at the operating-system level.
- Set a preferred language: Settings → Voice → Language can help speech recognition.
- Start a new call: changing voices during a call creates a new Voice call in the same chat.
- Test another network: unstable connections can cause pauses, dropped speech, or early call termination.
- Reduce background audio: use headphones; on iPhone, try Control Center → Mic Mode → Voice Isolation.
- Check the usage notice: reaching the rolling limit or maximum session/context limit can end a conversation.
Privacy controls to review before using Voice
OpenAI says audio clips from Live and Advanced Voice conversations are stored with the chat transcript for 30 days. Deleting the chat also schedules associated clips for deletion within 30 days, subject to stated security, safety, and legal exceptions. Archiving a chat does not delete it or its clips. With Standard, audio is deleted after transcription unless the user has chosen to share audio for model improvement. [2]
OpenAI says audio or video clips are not used for model training unless the user chooses to share them through Data Controls. Transcripts and other files may be used depending on the plan and settings when “Improve the model for everyone” is enabled. Review those settings before discussing private work, health, financial, legal, client, or family information. [2]
What GPT-Live-1.1 does not mean
- It does not guarantee that every account immediately sees every new voice.
- It does not make Voice answers automatically accurate; important claims still need verification.
- It does not add video or screen sharing to Live; those remain Advanced features for eligible mobile users.
- It does not make the transcript a verbatim record.
- It does not mean the GPT-Live-1.1 API is generally available merely because ChatGPT has the model; OpenAI said API access was coming soon on October 15.
- It does not remove plan, workspace, regional, parental-control, or rolling usage restrictions.
The original GPT-Live launch explains that the system uses full-duplex interaction: it can continuously listen and generate speech rather than waiting for rigid turns. It can also delegate deeper search or reasoning work to another model while maintaining the conversation. Those architectural ideas explain the experience, but they are not a guarantee that every interruption, accent, search, or transcription will work perfectly. [3]
Frequently asked questions
Is GPT-Live-1.1 available now?
OpenAI says it began rolling out GPT-Live-1.1 and GPT-Live-1.1 Quip in ChatGPT Voice on October 15, 2026. Voice and model availability can still vary by account, plan, region, workspace settings, and app version. [1] [2]
How do I get the Bubbie voice?
Open Settings → Voice, select Live, then open Voice and look for Bubbie. If it is missing after an app update, the rollout may not have reached your region or account. [2]
What is GPT-Live-1.1 Quip?
OpenAI describes Quip as the smaller version of GPT-Live-1.1. The current Voice guide lists limited Quip access for Free users and three hours for Go users over a rolling 24-hour period. [1] [2]
Does Live support screen sharing?
No. OpenAI’s current help page says Live does not support video or screen sharing. Eligible mobile subscribers can use those capabilities in Advanced Voice instead. [2]
Can I use GPT-Live-1.1 through the API?
OpenAI’s October 15 announcement said GPT-Live-1.1 was “coming soon to the API.” That is not the same as general availability. GPT-Live-1 already has API access, but developers should verify the current model catalog before selecting a 1.1 model identifier. [4] [3]
Are Voice transcripts exact?
No. OpenAI says a Voice transcript may not exactly match what the user or ChatGPT said, particularly with overlapping speech, background noise, or a fast conversation. [2]
Last checked: October 17, 2026. This is a source-based setup guide, not a hands-on review of every voice, plan, device, language, or region. Features, limits, prices, model names, and availability can change.
Sources
- OpenAI: ChatGPT release notes — GPT-Live-1.1 rollout and new voices
- OpenAI Help Center: ChatGPT Voice — modes, voices, limits, setup, privacy and troubleshooting
- OpenAI: Introducing GPT-Live — architecture, capabilities, safety and limitations
- OpenAI announcement: GPT-Live-1.1 rollout and API status