Consumer TechLOT 261119 · 3 min read
ChatGPT adds live dictation and removes 10-minute recording cap
OpenAI has removed ChatGPT’s 10-minute dictation cap and started a global rollout of live transcription for free and paid users across mobile, web and desktop.
Key takeaways
- OpenAI removed ChatGPT’s 10-minute dictation recording limit.
- Live Dictation is rolling out globally across free and paid plans on iOS, Android, web and desktop.
- The feature displays transcript text while the user is still speaking.
- OpenAI made the announcement in a Nov. 18 post on X.
- No new maximum length, latency, accuracy or supported-language list has been disclosed.

By Tom Whitfield
OpenAI has removed ChatGPT’s 10-minute recording cap and begun a global rollout of Live Dictation, which puts words on screen while the user is still speaking. The feature covers every ChatGPT plan, free and paid, on iOS, Android, the web and the desktop app. Availability depends on when the rollout reaches each account.
OpenAI made the case for the release in a Nov. 18 post on X: “Dictation in ChatGPT is getting faster and more reliable.” The company has not supplied latency or accuracy figures, nor has it identified a replacement maximum recording length.
What changes for the user?
Before this release, ChatGPT dictation transcribed a recording only after the user stopped speaking. Live Dictation shifts the process: text appears while speech continues, giving immediate visual feedback before the message is finished.
Users start the function through the microphone icon in the message box. Tapping or clicking the audio waveform shows or hides the live transcript, keeping the captured text visible without leaving the dictation flow.
The visible transcript gives users a way to check what ChatGPT has captured while they continue speaking. The practical takeaway is simple: users now see the transcript as it forms, not only after they stop.
Why did the 10-minute cap disappear?
OpenAI’s release notes state: “We’ve also removed the 10-minute recording limit.” Atty Eleti, who works on voice features at OpenAI, said on X that the change came with the streaming upgrade.
OpenAI did not specify a new ceiling. The previous limit no longer defines how long one continuous dictation session may run, but the company has left the outer boundary open.
The rollout policy covers free and paid plans, so the feature does not require a paid subscription, and OpenAI has announced no separate charge. The practical takeaway is that long dictation no longer faces the old ten-minute cutoff, but OpenAI has published no replacement boundary.
How is Live Dictation different from ChatGPT Voice?
Live Dictation is a speech-to-text input method for ChatGPT’s message box. ChatGPT Voice is separate: it is the spoken-conversation mode that OpenAI moved to its GPT-Live models in July.
One turns speech into visible text, while the other supports voice-based interaction with ChatGPT. The distinction makes this a change to message entry rather than a documented revision of the conversational voice mode.
OpenAI did not describe any changes to ChatGPT Voice or connect Live Dictation to the GPT-Live models. The practical takeaway is that Live Dictation changes message entry, while ChatGPT Voice remains a separate conversation feature.
What has OpenAI left unanswered?
OpenAI has not disclosed a new maximum recording length, latency, accuracy or supported-language list. That leaves a basic benchmark unavailable: users cannot judge from published figures how closely the display will track speech or how well it will perform in different languages.
Users can stop sharing dictation audio for model training through ChatGPT’s data controls. The rollout has broad plan and platform reach, but staggered account access means availability is not simultaneous.
OpenAI also has not identified which languages Live Dictation supports. The practical takeaway is that broad access is confirmed, while performance metrics and language coverage remain unspecified.
What should users and OpenAI watch next?
For users, the practical test is whether on-screen words keep pace with speech and remain accurate over long sessions. Those are precisely the measures OpenAI has not supplied, and they matter more than the feature’s broad initial reach.
Because OpenAI has announced no separate charge and includes free and paid plans, wider access will not distinguish those tiers. Going forward, the feature’s commercial value will depend on evidence of speed, accuracy and language coverage, which OpenAI has not yet supplied.
mlq.ai (Source)
More from Tom Whitfield
Show full bio
Staff writer covering companies, deals and strategy in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.
1,127 articles