Voxtral 3
Mistral's speech-to-text, text-to-speech and voice agent model family
Context window
Not published
Input / 1M tokens
Per audio minute (see note)
Output / 1M tokens
Per character for TTS (see note)
Provider
Mistral AI
Transcription is billed per audio minute: Voxtral Mini Transcribe 3 $0.0035, Voxtral Small Transcribe 3 $0.004, Voxtral Mini Transcribe Realtime 3 $0.008. Voxtral TTS 3 costs $13 per 1M characters. The voice agent API costs $0.04/min with Mistral Small 4 or $0.08/min with a larger model. · Data verified 2026-10-23
Voxtral 3 (October 20, 2026) is Mistral AI's third generation of speech models: two batch transcription models (Small and Mini), a realtime transcription model with sub-500ms latency, and a text-to-speech model with voice cloning, plus a voice agent API and SDK. Transcription covers 50+ languages including all 24 official EU languages, and transcription starts at $0.0035 per minute. Mistral offers it through its EU-hosted API or for self-deployment on private infrastructure.
Capability index
Relative estimates (0-100) to place this model against its peers, grounded in published benchmarks.
How to access it
Mistral API with model IDs voxtral-small-transcribe-3-0, voxtral-mini-transcribe-3-0, voxtral-mini-transcribe-realtime-3-0 and voxtral-mini-tts-3-0, or self-deployment via Mistral.
Strengths
- ✓Low transcription price from $0.0035 per audio minute
- ✓Realtime transcription with sub-500ms latency (Mistral-reported)
- ✓Speaker diarization and voice cloning
- ✓EU-hosted API or self-deployment
Best for developers who...
Benchmarks
| Benchmark | Score | Notes |
|---|---|---|
| Transcription WER | 4.1% | Voxtral Small and Mini Transcribe 3, Mistral-reported; Mistral lists Gemini 3.5 Transcribe at 4.0% on the same chart |
| Diarization cpWER | 18.4% | Voxtral Small Transcribe 3, Mistral-reported (Mini: 18.6%) |
Source: Mistral AI announcement
Compare Voxtral 3
Compare Voxtral 3 with any other model
Build a comparison →All model comparisons →