Voxtral Small Transcribe 3 · Voxtral Mini Transcribe 3
Audio transcription
Voxtral Transcribe 3 — Audio transcription
Voxtral Transcribe 3: Turn an uploaded audio or video file into a transcript. Choose a model, optionally add a language hint, and continue in Vizify with the same file and settings.
What should I upload? What will I receive?
Explore your idea with AI before generating.
Sign-in and access checks apply in Vizify.
Verified catalog facts
What Vizify exposes for this model family.
These facts are generated from the current public catalog, not copied from provider marketing.
Audio · Video
Text
Audio transcription
Reviewed 2026-11-01
Where this model fits.
Voxtral Transcribe 3: Turn an uploaded audio or video file into a transcript. Choose a model, optionally add a language hint, and continue in Vizify with the same file and settings.
What not to assume.
Speaker labels and word-level timing are retained when returned, but are not guaranteed on these channels. No live transcription or audio translation is exposed by this form.
The current balance check reserves usage for the maximum supported input duration. Final usage is reconciled after processing; shorter files can still need sufficient balance for that initial check.
What should I upload?
Upload one audio file or one video file, not both. A language hint is optional. Audio uploads are limited to 20 MiB and video uploads to 25 MiB; recordings can be up to 60 minutes.
What will I receive?
The task delivers a transcript with Markdown, SRT, WebVTT and JSON files. Review the recognized text before relying on it for captions, quotations or further editing. Voxtral Transcribe 3 offers Small and Mini. This entry selects Small; change to Mini in the model selector before submitting if that is the variant you want. Both accept the same file types, but their rates are separate. The integration requests segment timestamps and speaker labels; inspect the returned subtitles instead of assuming word-level timing.
Capability review
The form and delivery contract were reviewed against the Vizify integration on 2026-11-01. This is not a guarantee of channel availability, speed or output quality.
FAQ
Questions about Voxtral Transcribe 3 — Audio transcription
What should I upload?
Upload one audio file or one video file, not both. A language hint is optional. Audio uploads are limited to 20 MiB and video uploads to 25 MiB; recordings can be up to 60 minutes.
What will I receive?
The task delivers a transcript with Markdown, SRT, WebVTT and JSON files. Review the recognized text before relying on it for captions, quotations or further editing. Voxtral Transcribe 3 offers Small and Mini. This entry selects Small; change to Mini in the model selector before submitting if that is the variant you want. Both accept the same file types, but their rates are separate. The integration requests segment timestamps and speaker labels; inspect the returned subtitles instead of assuming word-level timing.
What are the current limits?
Speaker labels and word-level timing are retained when returned, but are not guaranteed on these channels. No live transcription or audio translation is exposed by this form.
How is usage estimated?
The current balance check reserves usage for the maximum supported input duration. Final usage is reconciled after processing; shorter files can still need sufficient balance for that initial check.