Voxtral Mini Transcribe 3 logo

Voxtral Mini Transcribe 3

mistral · Released Oct 2026

Conditional

Conditional. Voxtral Mini Transcribe 3 is the cheapest of Mistral's three Voxtral 3 transcription models at $0.0035 per minute, and the launch post describes diarization and context biasing in over 50 languages. The deciding caveat is that the model page labels it Public Preview and its third-place FLEURS ranking is Mistral's own claim. Suits teams transcribing recorded audio on a budget who can check accuracy on their own audio; pick Small for the flagship tier or Realtime for streaming.

Is it right for you?

Good for

  • Recorded-audio transcription at the lowest price of the three Voxtral 3 transcription models; the launch post's pricing table lists it at $0.0035/min and labels it "(batch)"
  • Multilingual work with speaker labels and custom vocabulary; the launch post says it supports multilingual transcription, diarization and context biasing in over 50 languages (Mistral's claim, untested by us)
  • Timestamped transcripts; the model page lists Timestamps and Transcriptions on /v1/audio/transcriptions

Not good for

  • Live streaming transcription; Mistral sells that as the separate Voxtral Mini Transcribe Realtime 3 at $0.008/min
  • Self-hosting from open weights; the pages we read state no open weights or licence, the model page shows a Premier badge, and its compact-design wording refers to deployment, not to downloadable weights
  • Anything that needs accuracy confirmed outside Mistral; the third-place FLEURS ranking appears on Mistral's own model page and we have not tested it
  • Production dependence on a settled status; the model page badges it Public Preview

How it performs by task

Transcription accuracy

Good

Mistral's model page says it ranks 3rd on the FLEURS multilingual transcription and English diarization benchmarks, ahead of Gemini 3.5 Transcribe; vendor claim, untested by us

Cost per audio minute

Very Good

$0.0035/min is the lowest price among the Voxtral 3 transcription rows in the launch post table; we did not compare other vendors

Speaker diarization

Good

The launch post and model description name diarization, but the model page feature list shows only Timestamps and Transcriptions; untested by us

Pricing

Input

$0.0035 / min of audio

Output

Billed per minute (single rate)

Context

Per-minute billing; 32k context per docs

View full pricing

Benchmarks

BenchmarkScoreSource
Rank on FLEURS multilingual transcription and English diarization (Mistral-reported)3rd Source

No verdict changes yet

The clock starts day one. Changes land here as our verdict evolves.

Verification log

  • Pricing· No changes

    Automated agent

  • Pricing· No changes

    Automated agent

  • Content· No changes

    Automated agent

How we evaluate