# Mistral open-sources Voxtral, a top-tier text-to-speech model

> Mistral released Voxtral, an open-weight TTS that runs on one GPU, in March 2026.

*Open weights, one GPU, aimed at ElevenLabs. The brief.*

By The FeaturedDaily Desk · FeaturedDaily
Canonical: https://featureddaily.com/news/mistral-voxtral-tts-brief

> **Key:** **The one-liner:** Mistral gave away a near-frontier AI voice you can run on one graphics card — pressure on the whole rent-the-voice industry.

**What happened.** In March 2026 Mistral released **Voxtral**, a 4-billion-parameter open-weight text-to-speech model that runs on a single 16GB GPU, clones a voice from a few seconds of audio, and supports nine languages.

**The claim.** Mistral's own blind tests preferred Voxtral over ElevenLabs Flash v2.5 roughly 63-70% of the time (vendor-run; attribute, don't assert).

> **Note:** **The catch.** Weights are CC BY-NC 4.0 — free for non-commercial use only; businesses pay via Mistral's API (~$0.016 per 1,000 characters). Still the standout open voice model months later.

## Key takeaways

- What: Voxtral, a 4B open-weight TTS that runs on a single 16GB GPU.
- Ability: clones a voice from seconds of audio; supports nine languages.
- Claim: Mistral's blind tests prefer it over ElevenLabs Flash v2.5 (63-70%) — vendor-run.
- Licence: CC BY-NC 4.0 (free non-commercial); commercial via Mistral's API.

## FAQ

### What is Voxtral?
Mistral's open-weight text-to-speech model, released March 2026 — 4B parameters, runs on a single consumer GPU, clones voices from seconds of audio, and supports nine languages.

### Is it free for business use?
No — the free downloadable weights are non-commercial (CC BY-NC 4.0). Commercial use runs through Mistral's API at around $0.016 per 1,000 characters.