ChatGPT’s dictation just got a big upgrade. On 18 November 2026, OpenAI rolled out Live Dictation: your words now appear on screen while you speak, and the old 10-minute recording limit is gone.
It is free on every ChatGPT plan and is rolling out worldwide on iPhone, Android, the web and the desktop app.
This guide shows how to use Live Dictation step by step, how it differs from Voice mode, what happens to your audio, and how to fix it when the mic button misbehaves.
- What: ChatGPT dictation now shows text live as you talk, instead of only after you stop.
- Limit: the 10-minute cap per recording is removed. OpenAI has not published a new maximum.
- Who: all plans, including Free. Rolling out gradually around the world.
- Where: ChatGPT on iOS, Android, web and desktop.
- How: tap the microphone in the message box, speak, tap the waveform to show or hide the live text, then edit and send.
- Privacy: audio is processed by OpenAI; it is only used for training if you turned on audio sharing.
What Changed
Dictation in ChatGPT is not new. You could already tap the mic, talk, and get your words typed into the message box. The difference was timing: ChatGPT recorded first and transcribed at the end, so you saw nothing until you stopped.
OpenAI announced the update in its ChatGPT release notes and on X, saying dictation is “getting faster and more reliable”. Here is what is new:
| Feature | Before 18 Nov | Now |
|---|---|---|
| When text appears | After you stop recording | Live, while you speak |
| Recording length | 10 minutes maximum | No 10-minute cap (no new limit published) |
| Speed and reliability | Long clips could fail at the end | Faster, more reliable, per OpenAI |
| Show or hide text | Not available | Tap the waveform to toggle the live transcript |
| Plans | All plans | All plans, including Free |
Atty Eleti, who works on voice at OpenAI, confirmed on X that the 10-minute limit was removed, according to 9to5Mac.
The live view matters more than it sounds. With the old method, a long ramble could fail at minute nine and you would lose everything. Now you can see mistakes as they happen and stop to fix them.
Who Gets It and Where
| Question | Answer |
|---|---|
| Plans | Free, Go, Plus, Pro, Business, Enterprise and Edu (OpenAI says all plans) |
| Platforms | iOS, Android, chatgpt.com in a browser, and the ChatGPT desktop app |
| Countries | Global rollout, arriving gradually |
| Cost | Free; no extra charge on any plan |
| Usage caps | None announced |
| Languages | Not listed in the announcement; ChatGPT dictation already handles many languages |
Not seeing it yet? The rollout is gradual. Update the ChatGPT app from the App Store or Google Play, or refresh chatgpt.com, and check again over the next few days.
Not sure which plan you are on or what each one adds? See my ChatGPT Free vs Go vs Plus vs Pro comparison.
How to Use Live Dictation
On iPhone or Android
- Update the app to the latest ChatGPT version.
- Open a chat and look at the message box at the bottom.
- Tap the microphone icon (not the round Voice mode button). Allow microphone access the first time.
- Start talking. Your words appear as you speak.
- Tap the waveform if you want to hide the live text and just talk, or tap it again to show it.
- Tap stop when you finish. The full transcript lands in the message box.
- Edit the text if needed, then tap send.
On the web or desktop app
- Open chatgpt.com or the ChatGPT desktop app and sign in.
- Click the microphone icon in the message box. Your browser may ask for microphone permission; choose Allow.
- Speak and watch the text appear. Click the waveform to show or hide it.
- Click stop, fix any words, and press Enter to send.
Nothing is sent until you press send. That is the main safety net: you always get to read and correct the transcript first.
Live Dictation vs Voice Mode vs Record Mode
ChatGPT now has three different ways to use your voice, and they are easy to mix up.
| Live Dictation | Voice mode | Record mode (beta) | |
|---|---|---|---|
| What it does | Types your speech into the message box | Spoken conversation; ChatGPT talks back | Records meetings or voice notes and summarises them |
| You control sending | Yes, edit then send | No, it replies as you talk | Creates notes after recording |
| Plans | All, including Free | All, with limits by plan | Plus, Pro, Business, Enterprise, Edu |
| Best for | Long prompts, emails, notes you want to edit | Hands-free chats, practice, quick questions | Meetings and lectures |
For a full walkthrough of the spoken mode, see my ChatGPT Voice mode guide.
Tips for Better Dictation
- Talk your whole prompt at once. With no 10-minute cap, you can explain a big task in one go, then tidy it up before sending.
- Say the structure out loud. Phrases like “first”, “second” and “new paragraph” help ChatGPT understand your intent, even if punctuation is imperfect.
- Hide the live text if it distracts you. Some people think better without watching words appear. Tap the waveform to hide it.
- Spell tricky names. Brand names, code terms and unusual names still trip up speech recognition. Fix them in the box before sending.
- Use a headset in noisy places. A close microphone gives far cleaner transcripts than a laptop mic across the room.
- Pair it with custom instructions. Tell ChatGPT once that your prompts are dictated and may have small errors. My custom instructions guide shows how.
Privacy: What Happens to Your Audio
Notebookcheck summarised OpenAI’s dictation FAQ. The key points:
- Audio goes to OpenAI. Transcription happens on OpenAI’s servers, not on your phone.
- Kept with the chat. Audio is stored as long as the chat stays in your history. Delete the chat and the audio is deleted within 30 days, except where OpenAI must keep it for security or legal reasons.
- Training on audio is opt-in. Your recordings are only used to train models if you turned on “Include your audio recordings” in data controls.
- The text is a normal message. Once sent, the transcript is treated like anything you type, so the “Improve the model for everyone” setting applies.
- Defaults differ by plan. On Free, Plus and Pro, “Improve the model for everyone” is on by default. Business, Enterprise and Edu data is excluded from training by default.
How to keep dictation out of training
- Open ChatGPT and go to Settings.
- Choose Data controls.
- Turn off Improve the model for everyone.
- Make sure Include your audio recordings is off.
- For extra privacy, use a Temporary chat for sensitive dictation.
Tip: don’t dictate passwords, card numbers or other people’s private details. Even with training off, the audio is still processed and stored with the chat.
ChatGPT vs Other Dictation Tools
| Tool | Cost | Where it works | Best for |
|---|---|---|---|
| ChatGPT Live Dictation | Free in ChatGPT | Inside ChatGPT only | Long prompts you will send to ChatGPT |
| Phone keyboard dictation (iOS, Gboard) | Free | Any app | Quick messages anywhere |
| Windows voice typing (Win + H) | Free | Any Windows app | Typing into documents and email |
| Gemini app voice input | Free | Inside Gemini | Prompts for Gemini |
| ElevenAI by ElevenLabs | Free preview | Meetings and dictation | Meeting notes with follow-ups |
The ChatGPT version’s main advantage is that it is tuned for prompts and now handles long, rambling input without a time limit. For dictating into other apps, your phone’s keyboard or Windows voice typing is still more convenient. For meetings, see my ElevenAI guide.
Troubleshooting
| Problem | Likely cause | Fix |
|---|---|---|
| Text only appears after I stop | Update not reached your account yet | Update the app, refresh the web page, wait a few days for the rollout |
| Microphone icon missing | Mic permission blocked, or an old app version | Allow microphone access in phone or browser settings; update ChatGPT |
| Voice mode opens instead | Tapped the round Voice button | Use the small microphone icon in the message box |
| Transcript has wrong words | Background noise or unusual names | Move closer to the mic, use a headset, correct names before sending |
| Recording stops early | Network drop or app sent to background | Keep the app open on a stable connection; send long dictation in parts if it keeps failing |
| “Something went wrong” error | Server or connection issue | Try again, check status.openai.com, see my ChatGPT not working fix guide |
My Take
This is a small update that will make a big difference for anyone who talks to ChatGPT more than they type. Seeing the words live removes the worry of losing a long recording, and dropping the 10-minute limit means you can explain a whole project out loud.
It is also free for everyone, which is rare for a voice feature. Just check your data controls first if you plan to dictate anything private.
FAQ
What is ChatGPT Live Dictation?
It is the updated dictation feature in ChatGPT that turns your speech into text on screen while you talk, so you can watch, edit and then send your prompt.
Is Live Dictation free?
Yes. OpenAI says it is available on all ChatGPT plans, including Free.
Is there still a time limit on ChatGPT dictation?
The 10-minute limit has been removed. OpenAI has not published a new maximum length.
When did Live Dictation launch?
OpenAI announced it on 18 November 2026, with a gradual global rollout.
Which devices support it?
ChatGPT on iPhone, Android, the web and the desktop app.
How is it different from Voice mode?
Dictation types your words into the message box for you to send. Voice mode is a spoken conversation where ChatGPT talks back.
Does OpenAI train on my dictation audio?
Only if you turned on “Include your audio recordings” in data controls. The text you send follows the “Improve the model for everyone” setting.
Why don’t I have Live Dictation yet?
The rollout is gradual. Update the app or refresh chatgpt.com and check again in a few days.
Sources
- OpenAI: ChatGPT release notes, 18 November 2026, and @ChatGPT post on X
- 9to5Mac: “ChatGPT app gains enhanced Live Dictation streaming without 10-minute limit”, 18 November 2026
- Notebookcheck: “ChatGPT Live Dictation removes the 10-minute limit for all plans” (includes OpenAI dictation FAQ details)
- tbreak: “ChatGPT Live Dictation types your prompts as you speak”




