Voxtral Mini Transcribe 2, version v26.02, is Mistral's efficient audio input model, pre-trained and optimized for transcription only. Its model page lists the API id voxtral-mini-2602, a Premier tag, a release date of February 4, 2026, and a price of $0.003 per minute. It is the current transcription model in Mistral's models list and replaces the older Voxtral Mini Transcribe 25.07.
Key Facts
| Item | Detail |
|---|---|
| API id | voxtral-mini-2602 |
| Version | v26.02 (Premier) |
| Released | February 4, 2026 |
| Price | $0.003 per minute |
| Endpoint | /v1/audio/transcriptions, with timestamps listed as a feature |
| Task | Transcription only |
The pages we checked on 2026-10-29 do not list supported languages, diarization, context biasing, or a maximum audio length, so we do not state them. Check the model page for the current feature list.
What It Replaced
The earlier Voxtral Mini Transcribe 25.07 (voxtral-mini-2507) is deprecated. Mistral's models list gives a deprecation date of February 27, 2026 and a retirement date of May 31, 2026, with Voxtral Mini Transcribe 2 as the replacement.
Use Cases
- Meeting, call, and interview transcription at a flat per-minute price.
- Subtitle workflows that use timestamps from the transcription endpoint.
- Batch transcription pipelines that feed text into search or summarization.
Limitation: it is transcription only. For generating speech, see Mistral's separate text-to-speech model, and for streaming use see the real-time variant.
Getting Started
- Create a Mistral API key.
- Call the
/v1/audio/transcriptionsendpoint withmodelset tovoxtral-mini-2602. - Compare output against your own audio, including accents and noisy samples, before scaling up.
FAQ
How much does Voxtral Mini Transcribe 2 cost?
$0.003 per minute, per Mistral's model page.
What is the API id?
voxtral-mini-2602.
Is the older 25.07 version still available?
No. It is deprecated, with retirement on May 31, 2026.
Is there a real-time version?
This catalog lists a separate real-time variant: Voxtral Mini Transcribe Realtime.
Alternatives
- Voxtral Mini Transcribe Realtime: the streaming variant.
- Mistral Voxtral TTS: Mistral's text-to-speech model.
- Qwen Audio 3.1: an audio model from Alibaba's Qwen line.
Conclusion
Voxtral Mini Transcribe 2 is Mistral's current low-cost transcription model. See the official model page and the models category for more.
Sources
- Websitehttps://docs.mistral.ai/models
- Social Mediahttps://x.com/MistralAI
Comments
No comments yet. Be the first to comment!