ChatGPT voice is not the “Advanced Voice Mode” you may remember. Since July 2026 the default is Live, powered by OpenAI’s full-duplex GPT-Live models, and on October 15, 2026 it moved to GPT-Live-1.1 with new voices.
This guide shows how to use ChatGPT voice on your phone, desktop and the web, which voices exist, the daily limits on each plan, how to use video and screen share, and how to fix common problems.
- Three modes: Live (default, full duplex), Advanced (has video and screen share) and Standard (turn by turn).
- Models: Plus and higher get GPT-Live-1.1. Free and Go get the smaller GPT-Live-1.1 Quip.
- Limits per rolling 24 hours: Go 3 hours, Plus 3 hours, Pro $100 15 hours, Pro $200 unlimited.
- Video and screen share: only in Advanced, on the iOS and Android apps, for Go, Plus and Pro.
- Privacy: Live and Advanced audio clips are kept with the chat for 30 days.
What Changed in ChatGPT Voice in 2026
OpenAI rebuilt ChatGPT voice this year. The old Advanced Voice Mode still exists, but it is no longer the default.
- July 8, 2026: GPT-Live launches as the new default voice. It is full duplex, so it listens while it talks.
- September 10, 2026: GPT-Live-1 reaches developers in the API at $0.05 per minute.
- September 23, 2026: Live voice gains plugin support, so it can use your connected apps.
- October 15, 2026: GPT-Live-1.1 and GPT-Live-1.1 Quip roll out, with new voices in six languages.
- October 22, 2026: the iOS app brings back the voice button right after a call ends.
Full duplex is the big change. Live can stop mid-sentence when you cut in, stay quiet during a pause, and give small “mhmm” style cues. In OpenAI’s own testing at launch, people preferred GPT-Live-1 over Advanced Voice in 75.7% of comparisons.
How Live answers hard questions
GPT-Live keeps talking while a stronger text model works in the background. Web searches and multi-step reasoning are handed off, then the answer is read back to you. At launch, Free users got the quick Instant level only, while paid users could pick higher reasoning levels.
Live vs Advanced vs Standard Voice
You pick the mode in Settings > Voice. Each one has a clear job.
| Mode | How it works | Video and screen share | Best for |
|---|---|---|---|
| Live (default) | Full duplex, interrupt any time | No | Natural talk, brainstorming, practice |
| Advanced | Older real-time voice | Yes, on iOS and Android | Showing things to the camera, custom GPTs |
| Standard | Turns speech into text, then replies | No | Weak connections |
How to Use ChatGPT Voice on Your Phone
The iOS and Android apps have the most complete voice features. Make sure the app is up to date first.
- Open the ChatGPT app and sign in.
- Tap the Voice button (the sound-wave icon) next to the message box.
- Allow microphone access the first time your phone asks.
- Start talking. You can interrupt at any point; Live will stop and listen.
- Mute or end the call with the buttons at the bottom of the screen.
- Read the transcript afterwards. The conversation is saved as a normal chat.
Tip: Set your main language in Settings > Voice > Language. It helps a lot if you speak with an accent or switch between languages.
How to Use ChatGPT Voice on Desktop and the Web
Voice in a normal chat runs on the web, iOS and Android. Voice in the desktop app covers macOS and Windows.
- Open chatgpt.com in a modern browser, or open the ChatGPT desktop app.
- Click the Voice button next to the message box.
- Allow the microphone in your browser or system privacy settings.
- Talk normally. Use a headset if you are in a noisy room, so Live does not hear itself.
- End the session to return to the text chat with the transcript.
Live voice is not available inside ChatGPT Work or Codex sessions, according to launch coverage. If you need a voice agent inside your own app, see my OpenAI Realtime API guide.
ChatGPT Voices and Languages
To change the voice, go to Settings > Voice and pick one. Nine existing ChatGPT voices were remastered for GPT-Live at launch.
The October 15 update added native-sounding voices for six languages:
- French: Aube, Rivage
- German: Sonna, Miko
- Hindi: Leher, Neer
- Japanese: Himari, Rindo
- Korean: Haetsal, Hanul
- Mexican Spanish: Tinta, Arrullo
Brazilian Portuguese voices (Viola and Rio) and a new English voice called Bubbie are also listed. Voice choice can vary by region and account.
Video and Screen Sharing
This is where people get confused. Live does not support video or screen sharing. You need Advanced voice for that.
- Go to Settings > Voice in the iOS or Android app.
- Choose Advanced as your voice mode.
- Start a voice chat with the Voice button.
- Tap the camera icon to show ChatGPT what is in front of you.
- Use the screen share option in the voice controls to share your phone screen.
- Switch back to Live in settings when you are done, if you prefer its feel.
- Who gets it: eligible Go, Plus and Pro users. Free has no voice with video.
- Separate limit: video has its own cap. When you hit it, the camera or screen feed may stop while the voice chat carries on.
ChatGPT Voice Limits per Plan
Live voice limits run over a rolling 24-hour window, not a calendar day. These figures come from OpenAI’s ChatGPT Voice help page as reported in mid-October 2026.
| Plan | Voice model | Live voice per rolling 24 hours | Video (Advanced) |
|---|---|---|---|
| Free | GPT-Live-1.1 Quip | Limited (no number published) | No |
| Go (about $8) | GPT-Live-1.1 Quip | 3 hours | Yes |
| Plus ($20) | GPT-Live-1.1 | 3 hours | Yes |
| Pro ($100) | GPT-Live-1.1 | 15 hours | Yes |
| Pro ($200) | GPT-Live-1.1 | Unlimited, with abuse guardrails | Yes |
| Business Standard | GPT-Live-1.1 | 3 hours, then 1.25 credits per minute | Varies |
| Business Premium | GPT-Live-1.1 | 15 hours, then 1.25 credits per minute | Varies |
- Pro $500: not listed separately in the voice limits we found.
- Single call length: one report says a single Live conversation caps at 2 hours.
- Enterprise: credit-based workspaces pay 1.25 credits per minute; usage-based USD workspaces pay $0.05 per minute.
For everything else each plan includes, see my ChatGPT plans comparison.
Privacy and Audio Storage
Voice chats are saved like text chats, with a few extras worth knowing.
- Audio clips: Live and Advanced audio, and Advanced video clips, are kept with your chat history for 30 days.
- Standard mode: reportedly deletes the audio after it is turned into text.
- Training: clips are not used to train models unless you opt in under Settings > Data Controls.
- Deleting: delete the chat to remove its transcript from your history.
- Memory: your memory settings apply to your whole account. See my ChatGPT memory guide to review what ChatGPT remembers.
Warning: do not read out passwords, card numbers or health IDs in a voice chat. Anything you say ends up in the transcript, and the audio is stored for 30 days.
Best Things to Do with ChatGPT Voice
Voice shines when your hands or eyes are busy, or when talking is simply faster than typing. Here are uses that suit Live well:
- Language practice: hold a real conversation in French, German, Hindi, Japanese, Korean or Spanish with a native-sounding voice.
- Interview prep: ask Live to play the interviewer and give feedback after each answer.
- Cooking and DIY: ask follow-up questions step by step while your hands are full.
- Thinking out loud: talk through a plan or a problem, then ask for a written summary in the chat.
- Quick facts on the move: Live can hand searches to a stronger model and show cards for weather, stocks and sports.
- Show and ask (Advanced): point the camera at a plant, a gadget or a math problem and ask what you are looking at.
Once a voice chat ends, the full transcript stays in your chat list. You can keep going by typing, or ask ChatGPT to turn the talk into notes, an email or a to-do list.
Tips for Better Voice Chats
- Use headphones: it stops Live from hearing its own voice and cutting itself off.
- Set the scene first: “You are my Spanish tutor. Correct my grammar after each answer.”
- Ask for short replies: “Keep answers under 20 seconds” makes the chat feel faster.
- Use plugins by voice: Live can use connected apps, but approvals must be tapped on screen, not spoken.
- Try live translation: ask Live to translate between two people speaking different languages.
Troubleshooting: Common Problems and Fixes
- No Voice button: update the app, sign out and back in, and check that voice is not turned off by your workspace admin.
- “Microphone not available”: allow mic access in iOS or Android settings, or in your browser’s site settings for chatgpt.com.
- No camera or screen share: switch to Advanced in Settings > Voice. Live does not offer video, and Free has no video at all.
- Voice cuts itself off: use headphones, or move away from speakers and background noise.
- Voice stopped after a few hours: you hit the rolling 24-hour limit. It frees up gradually as old usage drops out of the window.
- Laggy or robotic voice: switch from mobile data to Wi-Fi, or try Standard mode on a weak connection.
- Custom GPT voice feels different: custom GPTs use Advanced voice, not Live.
- Still broken: check the wider fixes in my ChatGPT not working guide.
Frequently Asked Questions
Is ChatGPT voice mode free?
Yes, Free users get limited Live voice on the smaller GPT-Live-1.1 Quip model. OpenAI does not publish an exact time limit for Free.
What is the difference between Live and Advanced voice?
Live is full duplex, so it listens while it talks and handles interruptions better. Advanced is the older mode, but it is the only one with video and screen sharing.
How long can I talk to ChatGPT on Plus?
Plus includes 3 hours of Live voice per rolling 24 hours on GPT-Live-1.1. Pro at $100 gets 15 hours, and Pro at $200 is unlimited with abuse guardrails.
Can ChatGPT voice see my screen?
Yes, but only in Advanced voice on the iOS and Android apps for Go, Plus and Pro. Live voice does not support video or screen sharing.
Does ChatGPT voice work on Windows and Mac?
Yes. Voice works on chatgpt.com in a browser, and the desktop app supports voice on macOS and Windows.
How do I change the ChatGPT voice?
Go to Settings > Voice and pick a new voice. You can also set your preferred language there.
Are my voice recordings saved?
Live and Advanced audio clips are kept with the chat for 30 days. They are not used for training unless you opt in under Data Controls.
