Voxtral Speech 3
#8 en Synthèse vocale (TTS)Mistral AI · v3 · depuis 20. Oktober 2026 · 2× · vu le 22 oct. 2026
41
Momentum
Voxtral Speech 3 (also called Voxtral Mini TTS 3 / voxtral-mini-tts-3-0) is the text-to-speech model of Mistral AI's Voxtral 3 product family, released on October 20, 2026 as a public preview via the Mistral API. According to Mistral's documentation it supports zero-shot and professional voice cloning, streaming with about 90 ms time-to-first-audio, and 34 emotion tags and 2 pace tags. Language figures conflict across sources (9 languages in the docs, 35+ languages and 50+ dialects in the announcement).
Historique du momentum
26.07.24.10.
Fonctionnalités
| Real-Time Streaming | Yes, streaming output with about 90 ms time-to-first-audio; batch and realtime via one API |
| Latency | about 90 ms time-to-first-audio when streaming |
| Platform | Mistral API (EU, US and global endpoints), Mistral Studio; enterprise self-hosting per announcement |
| Price | $13 per 1M characters (batch and realtime, same price) |
| Release Date | October 20, 2026 (public preview) |
| Languages | 9 languages per Mistral docs (announcement: 35+ languages, 50+ dialects) |
| Voice Cloning | Zero-shot and professional voice cloning (Pro Voice from about 30 min audio) |