Mistral's Voxtral 3 Voice Suite Launches: Third in Transcription, Tops TTS Speed
Mistral released the Voxtral 3 voice family on October 21, comprising one TTS model and three STT (speech-to-text) models. Artificial Analysis promptly benchmarked them and updated its voice leaderboards: transcription results are impressive, TTS wins on speed and price, but pronunciation accuracy lags.
Confirmed
- Voxtral Small Transcribe 3 ranks 3rd on the non-streaming transcription leaderboard with a 2.1% AA-WER Index, ahead of ElevenLabs Scribe v2 (2.2%) and Grok Voice.
- The streaming model Voxtral Mini Transcribe Realtime 3 offers two latency settings, both landing in the upper half of the streaming STT leaderboard: the 480ms latency setting achieves a 3.6% WER.
- Voxtral 3 TTS generates at 112.3 characters per second, far outpacing Eleven v4 Turbo to top the speed chart; it is priced at one-third of Eleven v4 Turbo.
- On the Pronunciation Robustness benchmark (which tests how accurately TTS handles challenging text, scored by human reviewers), Voxtral 3 TTS scored 84.4% overall, ranking 13th out of 32 models—exact sequence fidelity is its weak spot.
- Artificial Analysis also released updated TTS/STT leaderboards and Speech Arena voting data.
Why it matters
- Voxtral 3 enters the voice market with low prices and high speed, going straight after ElevenLabs' flagship lineup and giving developers a more cost-effective option.
- Cracking the top three in transcription shows Mistral has moved from text-only models into first-tier competition in speech multimodality; but its lower pronunciation robustness ranking suggests a gap on hard text, so choice depends on the use case.
2026-10-21 ~ 2026-10-21 · 6 related posts
Primary sources
- Mistral's Voxtral 3 speech models land #3 on STT leaderboard at $13/1M chars — ArtificialAnlys ·
- Voxtral Small Transcribe 3 Takes #3 on STT Leaderboard at 2.1% WER, Edging Scribe v2 — ArtificialAnlys ·
- Voxtral 3 TTS Tops Speed Charts at 112 chars/sec, Undercuts Eleven v4 Turbo by 3x on Price — ArtificialAnlys ·
- Voxtral 3 TTS Scores 84.4% on Pronunciation Robustness, Ranks #13 of 32 Models — ArtificialAnlys · 2026-10-21
- [source] Voxtral Small Transcribe 3 Takes #3 on STT Leaderboard at 2.1% WER, Edging Scribe v2 — ArtificialAnlys · 2026-10-21
- Voxtral Mini Transcribe Realtime 3 Hits Top-10 Streaming STT at 480ms and 960ms Latency — ArtificialAnlys · 2026-10-21
- [source] Voxtral 3 TTS Tops Speed Charts at 112 chars/sec, Undercuts Eleven v4 Turbo by 3x on Price — ArtificialAnlys · 2026-10-21
- [source] Mistral's Voxtral 3 speech models land #3 on STT leaderboard at $13/1M chars — ArtificialAnlys · 2026-10-21
1 near-duplicate retellings: ArtificialAnlys