Takeaway: On November 18, 2026, OpenAI’s ChatGPT release notes introduced Live Dictation. The company says your speech turns into text as you talk, so you spend less time waiting for a message to finish transcribing, and that the previous 10-minute recording limit has been removed. Live Dictation is rolling out globally to all ChatGPT plans on iOS, Android, web, and the desktop app. It is a dictation feature, not a change to ChatGPT Voice, and it is separate from the already announced retirement of Advanced and Standard Voice on December 11, 2026.
The details below come from those November 18 release notes, OpenAI’s Voice Dictation FAQ, and the official ChatGPT Voice help article, which distinguishes dictation from a live voice conversation. OpenAI has not published a price, a language list, or availability exceptions for Live Dictation beyond the rollout statement in the release notes, and this article does not add any.
What Live Dictation does
OpenAI describes Live Dictation as a faster, more reliable way to use dictation in ChatGPT. With the new behavior, speech is turned into text as you talk rather than after you finish a recording. You can tap or click the waveform to show or hide text as you go. The company also says it has removed the 10-minute recording limit that previously applied to dictation.
The Voice Dictation FAQ explains the underlying idea in more general terms: when you press the microphone icon to record an audio message, the recorded audio is sent to OpenAI’s models to be transcribed, and the transcription is returned as text that you can edit before sending as a user message. Live Dictation keeps that review-and-edit step, while changing how the text appears during the recording itself.
Dictation is not ChatGPT Voice
OpenAI’s Voice help article draws a clear line between the two tools. ChatGPT Voice is for a spoken, back-and-forth conversation in which ChatGPT also replies out loud. If you want to turn a single recording into editable text instead, the same article says to use ChatGPT Dictation. In its FAQ, OpenAI puts it this way: use Voice for a live conversation or to talk through ideas; use dictation when you want to record a prompt, review and edit the transcription, and then send it as text.
That distinction matters this month, because OpenAI is also retiring Advanced and Standard Voice from ChatGPT Voice on December 11, 2026, and moving remaining users to Live voice powered by GPT-Live. That is a change to the conversational voice product. Live Dictation is a November 18 update to the microphone-to-text composer flow. The similar “Live” label is easy to mix up, but the release notes present Live Dictation as a dictation improvement, not as a replacement for Voice, and not as part of the December 11 retirement.
Where it is rolling out
The November 18 notes say Live Dictation is rolling out globally to all ChatGPT plans on iOS, Android, web, and the desktop app. OpenAI does not list excluded countries, languages, or plan-by-plan exceptions in that entry. Until the company publishes more detail, the accurate statement is the official one: a global rollout across those clients and all ChatGPT plans.
Because it is described as a rollout, some accounts may see the new live transcription sooner than others. Official notes do not give a completion date. They also do not say whether older, non-live dictation remains available in parallel once the new behavior arrives.
What OpenAI says about audio retention and training
The Voice Dictation FAQ covers how dictation audio is handled. Audio from dictation is retained for as long as the chat is part of your chat history. When you delete the chat, OpenAI says it will also delete the associated audio clip within 30 days, unless it needs to keep the clip for security or legal reasons, or unless you previously shared your audio to train models and the clip was already disassociated from your account.
On training, OpenAI says that if you are a consumer user and have chosen to share audio to improve its models, it may train on audio from dictation. By default, the company says it does not train on any inputs or outputs from its products for business users, and it points to its Enterprise Privacy page for how business data is handled.
Users who want to stop sharing can do so in ChatGPT’s data controls. The FAQ says to disable the “Include your audio recordings” toggle, or to turn off “Improve the model for everyone” entirely. That change, OpenAI notes, stops sharing all audio across dictation and voice chat — another reminder that the two features are related in privacy settings even though they are different products in the interface.
None of this is a pricing or accounting-workflow change. Live Dictation sits in the composer, alongside everyday prompting, including business features such as the recently added ChatGPT for Accounting workflows. Those are separate products.
Key facts
| Item | What official sources say |
|---|---|
| Date | November 18, 2026, in the ChatGPT release notes. |
| Feature | Live Dictation: speech turns into text as you talk. You can tap or click the waveform to show or hide text as you go. |
| Recording limit | OpenAI says it has removed the previous 10-minute recording limit. |
| Availability | Rolling out globally to all ChatGPT plans on iOS, Android, web, and the desktop app. |
| How it differs from Voice | Dictation records a prompt, returns editable text, and lets you send it as a message. ChatGPT Voice is a spoken conversation with a spoken reply. |
| December 11 Voice change | Advanced and Standard Voice still retire on December 11, 2026. That retirement is a Voice-mode change, not the Live Dictation rollout. |
| Languages, price, exceptions | OpenAI did not publish a language list, a price, or availability exceptions in the official notes. |
| Audio retention | Dictation audio is kept while the chat remains in your history, then deleted within 30 days after the chat is deleted, with the official security, legal, and previously shared-for-training exceptions. |
FAQ
What is ChatGPT Live Dictation?
It is a dictation update announced in OpenAI’s November 18, 2026 ChatGPT release notes. With Live Dictation, your speech turns into text as you talk, and you can tap or click the waveform to show or hide that text as you go. You can still edit the transcription before sending it as a message.
Is the 10-minute recording limit still in place?
No. OpenAI says it has removed the 10-minute recording limit. The official notes do not describe a new time cap to replace it.
Which plans and apps get Live Dictation?
The release notes say Live Dictation is rolling out globally to all ChatGPT plans on iOS, Android, web, and the desktop app. OpenAI has not listed plan exceptions, excluded regions, or a rollout end date in that announcement.
Is Live Dictation the same as ChatGPT Voice or GPT-Live?
No. OpenAI says ChatGPT Voice is for a live, back-and-forth spoken conversation, while dictation is for recording a prompt, reviewing the transcription, and sending it as text. Live Dictation is an update to that dictation flow, not a new voice-conversation mode.
Does this change the December 11 Advanced and Standard Voice retirement?
No. OpenAI is still retiring Advanced and Standard Voice from ChatGPT Voice on December 11, 2026, with remaining users moving to Live voice. Live Dictation, announced on November 18, is a separate composer feature and does not replace that Voice-mode change.
Which languages does Live Dictation support?
OpenAI’s November 18 release notes and the Voice Dictation FAQ do not publish a language list for Live Dictation. This article does not infer one.
Does OpenAI keep the audio from dictation?
Yes, while the chat remains in your history. After you delete the chat, OpenAI says it deletes the associated audio clip within 30 days, unless it needs to keep the clip for security or legal reasons, or unless you previously shared the audio to train models and it was already disassociated from your account.
Does OpenAI train on dictation audio?
For consumer users who have chosen to share audio to improve OpenAI’s models, the company says it may train on audio from dictation. By default, it does not train on inputs or outputs from its products for business users. You can stop sharing in data controls by turning off “Include your audio recordings” or “Improve the model for everyone,” which stops sharing audio from both dictation and voice chat.

Leave a Reply
You must be logged in to post a comment.