Mistral's Voxtral 3 speech models: #3 STT accuracy, fastest TTS at $13/1M chars
ArtificialAnlys · x · 2026-10-21
Mistral released Voxtral 3: one TTS model and three speech-to-text models. Artificial Analysis evaluated them all.
- STT: Voxtral Small Transcribe 3 ranks #3 on the AA leaderboard at 2.1% AA-WER, ahead of ElevenLabs Scribe v2 (2.2%) and Grok Voice Transcribe 2.0 (2.3%); Mini Transcribe 3 lands #5 at 2.2%
- Streaming STT: Mini Transcribe Realtime 3 scores 3.3% WER at 960ms (#6 of 46) and 3.6% at 480ms, down from 5.2% for the previous version
- TTS: Voxtral 3 TTS ranks #12 in both arenas (Provider Voice Elo 1,218 of 98; Controlled Voice 1,068 of 43); pronunciation robustness 84.4%, #13
- Pricing: $13/1M characters — below Gemini 3.8 Flash TTS ($16.5) and Eleven v4 Turbo ($40), above Simba 3.2 ($6.6)
Related event: Mistral Launches Voxtral 3 Speech Models and Voice Agents API(10 posts)→
More from Multimodal
- What keeps AI videos watchable: hooks, consistency and a payoff — aftahi_ai · 2026-10-21
- MORPHJUTSU skill in Flova transfers any video's motion to your character — aftahi_ai · 2026-10-21
- NVIDIA open-sources MotionBricks: 350K motion skills at 15,000 FPS, 2ms latency — aftahi_ai · 2026-10-21
- Runway joins Claude Marketplace, letting enterprises spend Anthropic budget on video generation — runwayml · 2026-10-21
- Surreal AI football scenes: a pitch in a tree grid and dribbling past flowers — gizakdag · 2026-10-21
- Seedance 2.5 demo shows striking realism, prompt included — aftahi_ai · 2026-10-21