Technology
Mistral's free TTS aims straight at ElevenLabs
Open weights versus a rented-voice business model. Pick your side.
Mistral's open-weight Voxtral TTS (March 2026) targets ElevenLabs' rent-the-voice model.
Released March 2026: a 4B open-weight TTS that runs on a single 16GB GPU, clones a voice from seconds of audio, nine languages, preset voices. Mistral's own blind tests put it ahead of ElevenLabs Flash v2.5 63-70% of the time. Vendor numbers, obviously — but reviewers didn't laugh them off, which is the part that should sting.
The model-vs-model fight is the distraction
The whole premium-voice industry sells access: proprietary, API-only, you rent the voice. Voxtral attacks the structure, not the spec sheet — run it yourself, on your own servers, in a hospital or a bank where you can't ship audio to someone's cloud. When a 'good enough' open model exists, the incumbent's pricing power is the thing that quietly erodes.
Still, the direction is one-way. Mistral didn't have to win the voice market to hurt it — it just had to prove a near-frontier voice can be open and tiny. It did, months ago, and the option hasn't gone away. Rent-the-voice still works; it just has a clock on it now.
Sources
- Mistral AI just released a text-to-speech model it says beats ElevenLabs — and it's giving away the weights for free — VentureBeat, 26 March 2026
- Mistral releases an open-weights 'speaking' AI model with Voxtral TTS — SiliconANGLE, 26 March 2026