OpenAI adds live transcription to ChatGPT and removes its 10-minute dictation limit
Live Dictation is rolling out across all plans and platforms. Users can hide the running transcript, while OpenAI’s speed and reliability claims remain unquantified.
Loading page…
Live Dictation is rolling out across all plans and platforms. Users can hide the running transcript, while OpenAI’s speed and reliability claims remain unquantified.
Listen to this story
The update makes spoken input easier to monitor and removes a constraint on longer notes, but it does not establish a new maximum recording length or quantify performance. OpenAI announced Live Dictation on November 18, 2026, with streaming text users can reveal or hide while speaking. It is rolling out globally across all ChatGPT plans on iOS, Android, web, and desktop; availability may vary during rollout. OpenAI calls it faster and more reliable but supplied no latency or accuracy benchmarks.
Users can tap or click the audio waveform to show or hide the transcript during dictation.
The previous 10-minute cap is gone, but OpenAI did not state a replacement maximum, so recordings are not guaranteed to be unlimited.
The rollout covers all ChatGPT plans on iOS, Android, the web, and the desktop app, though not every account may have access yet.
People dictating into ChatGPT can now see their words appear while they speak, rather than wait for a finished transcription. OpenAI announced Live Dictation on November 18, 2026, alongside removal of the previous 10-minute recording limit. The update changes both when users can read their spoken input and how long they can keep talking.
Previously, ChatGPT users had to finish dictating and wait for processing before seeing the resulting text, according to Digital Trends. That separated speaking from checking what the app had heard. With the new streaming transcription, those two steps can overlap: the words become visible before the recording is over.
The practical benefit is earlier feedback. Digital Trends notes that seeing the transcript unfold makes it easier to spot misheard words or adjust phrasing mid-thought. That is a change in visibility, not a published measure of transcription accuracy: users get to inspect the text sooner, rather than simply trust that the recording was understood.
Reading along is optional. OpenAI says users can tap or click the audio waveform to show or hide the text during dictation. The control accommodates a different preference: speaking without a running transcript competing for attention. Users can switch its visibility while they are talking, rather than commit to one presentation throughout the recording.
OpenAI describes the update as faster and more reliable, saying users will spend less time waiting for a message to finish transcribing. But it published no specific latency benchmarks or accuracy metrics alongside the announcement, as Digital Trends notes. There is therefore no announced numerical comparison for how much faster transcription is or how often it gets words right.
Access is not restricted to a particular subscription or device family. OpenAI says Live Dictation is rolling out globally to all ChatGPT plans on iOS, Android, the web and the desktop app. That is a rollout announcement, rather than a promise that every account already has the updated experience. As it reaches an account, users can start dictating through the microphone icon.
For users considering longer spoken notes, audio-sharing preferences are managed separately in ChatGPT’s data controls. Digital Trends reports that users can disable inclusion of audio recordings or turn off model-improvement sharing. Those settings—not whether the live transcript is visible—are the controls to review when deciding how dictation audio is shared.
Loading discussion...
Join the conversation
Share whether seeing your words would sharpen your thoughts or break your flow.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.