Voxtral Mini Transcribe 3
Voxtral Mini Transcribe 3 converts speech to text and identifies speaker segments. It entered public preview in October 2026. Mistral positions Mini as the compact transcription option. The documented endpoint provides transcription and timestamps.
Preview
Use the transcription endpoint
The model page lists a 32k context and links to /v1/audio/transcriptions. It reports multilingual transcription and English diarization evaluations. Those vendor benchmark claims should be kept separate from accuracy on a particular accent, microphone or specialist vocabulary.Voxtral Mini Transcribe 3 model page
Evaluate the audio workflow
Practical guidance. Compare transcript errors and speaker boundaries on actual recordings. Include overlapping voices and uncommon names. The right compact model is the one that produces usable text within the required response time. Ranking on a broad language benchmark does not answer that application-specific question.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Preview |
| Launch date | |
| Access | API |
| Model identifiers | voxtral-mini-transcribe-3-0 |
| Context window | 32,000 tokens The model page advertises 32k context. This does not specify a maximum recording duration. Voxtral Mini Transcribe 3 model page |
| Initial release stage | Public preview.Voxtral Mini Transcribe 3 model page |
| API identifier | voxtral-mini-transcribe-3-0Voxtral Mini Transcribe 3 model page |
| Access | Premier API service; no public weights confirmed for this releaseVoxtral Mini Transcribe 3 model page |
Related models
- Voxtral Small Transcribe 3
Mistral AI · Voxtral
A preview speech model that combines transcription and diarization with speech translation and audio question answering.
Launch date PreviewAPI A streaming transcription model with turn detection, speaker diarization and configurable latency over WebSocket.
Launch date PreviewAPI