ChatGPT Realtime Voice 2026: 3 hours a day on Plus, setup and fixes
Updated: 2 minutes ago
Updated: November 11, 2026
ChatGPT Voice now runs on GPT-Live-1.1, and Plus subscribers get up to 3 hours of it in any rolling 24-hour period, while Free users get limited access to the smaller GPT-Live-1.1 Quip model and the $200 Pro plan has no fixed cap beyond abuse guardrails. The realtime conversation mode, called Live in the app, listens and speaks at the same time, so you can interrupt it, change direction mid-answer and keep talking while it searches the web, reads an image you added or calls a plugin.
The guide below covers the limit for every plan in a single table, the three voice modes that still coexist in Settings until OpenAI's planned December 11, 2026 retirement of Advanced and Standard voice, the steps to start Voice on iOS, Android, the web and the desktop app, what Business and Enterprise workspaces pay per minute, a troubleshooting table and the 30-day retention rule for audio clips. All plan details were checked against OpenAI's ChatGPT Voice help article and ChatGPT release notes on November 11, 2026.
··········
CHATGPT VOICE LIMITS BY PLAN IN NOVEMBER 2026.
OpenAI measures Voice in Chat as time used inside a rolling 24-hour window, and the model you talk to depends on the plan. Since the September 9, 2026 update, Plus and Pro users no longer drop to a smaller model when they hit a limit; the call ends and you can continue in text.
........
Plan | Voice model | Voice in Chat limit (rolling 24 hours) |
Free | GPT-Live-1.1 Quip | Limited access, may change |
Go | GPT-Live-1.1 Quip | Up to 3 hours |
Plus | GPT-Live-1.1 | Up to 3 hours |
Pro ($100 a month) | GPT-Live-1.1 | Up to 15 hours |
Pro ($200 a month) | GPT-Live-1.1 | Unlimited, subject to abuse guardrails |
Business Standard | GPT-Live-1.1 | 3 hours, then 1.25 credits per minute |
Business Premium | GPT-Live-1.1 | 15 hours, then 1.25 credits per minute |
Enterprise, Edu (credit pricing) | GPT-Live-1.1 | 1.25 credits per minute |
Enterprise (usage-based USD) | GPT-Live-1.1 | $0.05 per minute |
Legacy Enterprise and Edu, K-12 Teachers | GPT-Live-1.1 | 3 hours |
........
OpenAI's pricing page describes the same ladder in plainer words: "Limited voice chats" on Free, "More voice chats" on Go, expanded Voice on Plus and unlimited Voice on Pro. Voice used inside Work counts toward the normal Work allowance, and the desktop app has its own pricing and limits. For the wider picture of message caps, Codex pools and tool quotas, see our guide to ChatGPT usage limits by plan.
··········
GPT-LIVE, ADVANCED AND STANDARD: THE THREE CHATGPT VOICE MODES.
The Voice settings page currently lists three modes. Live is the default realtime experience introduced on July 8, 2026 with GPT-Live-1 and upgraded to GPT-Live-1.1 and GPT-Live-1.1 Quip on October 15, 2026, with more natural accents across languages. Advanced is the previous realtime system, and Standard is the older turn-by-turn mode that transcribes your speech first and then generates a reply.
........
Mode | How it works | Video and screen sharing | Status |
Live (GPT-Live-1.1 or Quip) | Listens and speaks at once, uses search, memory, plugins, images | Not supported | Default |
Advanced | Previous realtime voice | Supported on iOS and Android for eligible subscribers | Retirement announced for December 11, 2026 |
Standard | Turn by turn, speech transcribed before the reply | Not supported | Retirement announced for December 11, 2026 |
........
OpenAI said in a post on its official Newsroom account on X that Advanced and Standard voice will be retired from ChatGPT Voice on December 11, 2026, and that voice usage has nearly doubled since GPT-Live arrived in July. When we checked on November 11, the help center article still described all three modes and had not yet documented the retirement or said what will replace live camera and screen sharing, which today work only in Advanced. Users who rely on pointing the phone camera at an object or sharing a screen during a call should plan for that gap. The full announcement is covered in our report on the December 11 voice retirement.
··········
HOW TO TURN ON CHATGPT REALTIME VOICE ON IOS, ANDROID, WEB AND DESKTOP.
On iPhone and Android, open a chat and tap the Voice icon in the message bar, allow microphone access when the system asks, choose a voice the first time and start speaking. On the web, go to ChatGPT.com, select the Voice icon in the prompt window, allow the browser to use the microphone and speak. The microphone control mutes and unmutes you, and the exit control ends the call. You can have only one Voice conversation at a time.
To pick the mode, open Settings, then Voice. The same menu holds the voice choice, the Language setting (useful when Voice detects the wrong language), Background conversations, which keeps a call going while you use other apps or lock the phone, and Start with Voice, which opens Voice automatically in a new or empty chat on supported mobile versions. On iPhone, Live content can also appear on the Lock Screen and in the Dynamic Island through Live Activities, a feature OpenAI added on August 31, 2026.
In the ChatGPT desktop app for macOS and Windows, Voice works inside Chat, Work and Codex. You select Start voice chat in an existing task or Start new voice chat in a new one, grant microphone access, choose a voice and, on macOS, review screen context, which lets you say "Take a look at this" so Voice captures the frontmost window. A hotkey can be set under Settings, then Voice, then Voice chat hotkey. OpenAI's desktop documentation lists this experience for Plus, Pro, Business, Edu and Enterprise, with Enterprise and Edu starting on a two-week early-access period.
··········
CHATGPT VOICE FEATURES: VOICES, PLUGINS, FILES, CARPLAY AND BACKGROUND CALLS.
The standard voice set has nine options: Arbor, Breeze, Cove, Ember, Juniper, Maple, Sol, Spruce and Vale. Live adds language-specific voices for Brazilian Portuguese (Viola, Rio), French (Aube, Rivage), German (Sonna, Miko), Hindi (Leher, Neer), Japanese (Himari, Rindō), Korean (Haetsal, Hanul) and Mexican Spanish (Tinta, Arrullo), plus Bubbie in English. Availability varies by region and account, and switching voice during a call starts a new call in the same chat. Preset personalities do not apply to Live, although you can ask for a different tone, pace or style, and OpenAI offers no precise playback-speed control.
Live can use plugins on web, iOS and Android, including on Free and Go within the plugins those plans support. Approvals still happen on screen, since spoken approval is not supported, and workspace restrictions apply as in text chat. You can add text and images to the same chat through the add button and attach supported files manually, but Live cannot search or pull files from your ChatGPT Library. Since August 7, 2026, GPT-Live also works with file uploads and Projects, and Voice is available in Work on web and mobile, where an unfinished task can carry on as text after you hang up.
On supported iPhones, ChatGPT works in CarPlay, where you can start Voice, continue a recent or pinned chat or open a project chat, with an auto-start option that appears after the first use in the car. Background calls end when you end them, force-close the app, reach a usage limit or hit the maximum session length, and on mobile ChatGPT may end a background call that looks inactive if you do not respond to its check. For a closer look at how GPT-Live handles turn-taking, research during a call and everyday use cases, see our GPT-Live for ChatGPT Voice explainer.
··········
[[ADPLUS_300x250_1]]
CHATGPT VOICE IN BUSINESS AND ENTERPRISE: CREDITS AND PER-MINUTE COSTS.
Workspaces pay for Voice differently from personal plans. Business Standard includes 3 hours and Business Premium 15 hours per rolling 24 hours, after which extra usage is billed at 1.25 credits per minute. Enterprise, Edu and Clinicians workspaces on credit pricing pay 1.25 credits per minute from the first minute, and Enterprise workspaces on usage-based USD billing pay $0.05 per minute.
........
Scenario | Minutes | Usage-based Enterprise ($0.05/min) | Credit pricing (1.25 credits/min) |
One extra hour in a day | 60 | $3.00 | 75 credits |
30 minutes past the Business Standard allowance | 30 | Not applicable | 37.5 credits |
20-minute call every workday for 21 days | 420 | $21.00 per user | 525 credits per user |
........
The arithmetic is minutes multiplied by the published rate. OpenAI's Voice help article does not state a dollar value per credit, so the credit column cannot be converted into dollars from that page alone. At $0.05 per minute, a team of 50 people each using 420 minutes a month would spend about $1,050 a month on Voice under usage-based billing, which is the figure an administrator should compare against the cost of moving heavy users to a plan with an included allowance.
··········
CHATGPT VOICE TROUBLESHOOTING: COMMON PROBLEMS AND FIXES.
Most failures trace back to permissions, noise, limits or a second call already running. The table pairs the symptoms OpenAI documents with its recommended fixes.
........
Problem | Likely cause | Fix |
Voice button missing | Plan, rollout or workspace setting | Check your plan and workspace policy, update the app |
Call will not start | Microphone permission denied | Allow microphone access in iOS, Android or browser settings |
Call ends suddenly | Usage limit, maximum session length or context limit | Continue in text or start a new Voice call |
Live answers background chatter | Designed for one-on-one talk | Use headphones, move somewhere quieter, turn on Voice Isolation in iPhone Control Center |
Live replies while you think aloud | Long pauses read as a finished turn | Ask Live at the start to wait until you say you are ready |
Wrong language detected | Automatic detection | Set Settings, Voice, Language, or ask it to switch |
Wrong date or time in answers | Device or browser time zone | Check the time zone or state the exact date and place |
Start voice chat missing on desktop | Old app or Codex host | Update the desktop app and the Codex host |
No camera or screen sharing | Live does not support them | Use Advanced on iOS or Android until December 11 |
........
When Voice is the wrong tool, Dictation is the alternative. Dictation records a prompt, lets you edit the transcription and sends it as text, and OpenAI rolled out a new speech-to-text model for it on June 26, 2026 that it says cut word error rate by at least 10% for the top tested languages in internal evaluations.
··········
CHATGPT VOICE PRIVACY: TRANSCRIPTS, AUDIO CLIPS AND 30-DAY RETENTION.
After a call, ChatGPT adds a transcript to the chat, and Live shows its replies as text while it speaks. OpenAI warns that transcripts are not verbatim and can differ from what was said, especially with overlapping speech or background noise. Audio clips from Live and Advanced calls, and video clips from Advanced, are stored with the transcript for 30 days. Deleting the chat deletes those clips within 30 days unless they must be kept for security, safety or legal reasons, and archiving a chat does not delete them. Standard voice deletes the audio after transcription, even when transcription fails, unless you chose to share it for training.
Clips are used for model training only if you opt in through "Include your audio recordings" or "Include your video recordings". Transcripts follow the "Improve the model for everyone" setting, depending on plan. Free, Plus and Pro users in a personal workspace manage both under Settings, then Data Controls, and turning sharing off affects only new clips.
··········
CHATGPT REALTIME VOICE BEFORE THE DECEMBER 11 SWITCH.
ChatGPT Voice in November 2026 is a GPT-Live product with clear daily allowances: limited Quip access on Free, 3 hours on Go and Plus, 15 hours on the $100 Pro plan and no fixed cap on the $200 plan, with workspaces billed at 1.25 credits or $0.05 per minute beyond their allowances. Setup takes a tap on the Voice icon and a microphone permission on every platform, and the desktop app extends the same conversation into Work and Codex.
The open point is the December 11 retirement. OpenAI has announced it on its official X account, while its help center still lists Advanced and Standard and has not explained how camera and screen sharing will work once Advanced is gone. Until OpenAI documents that change, anyone who depends on video during calls should keep Advanced selected for those tasks and expect to check the Voice settings again after December 11.
··········
FOLLOW US FOR MORE.
·····
DATA STUDIOS
·····
datastudios.org
