M
Voxtral TTS 3
Voxtral TTS 3, a text-to-speech model from Mistral.
speechModelOverride: { model: "voxtral-tts-3" }
Mistral·Released October 2026Pricing
- List price
- $13 / 1M characters
- Per 1M characters
- $13.00
Specifications
- Context window
- 5,000 tokens
- Released
- October 2026
- Model string
- voxtral-mini-tts-3-0
- Capabilities
- MultilingualLow cost
Parameters
Model options, passed in
speechModelOverride.config.| Parameter | Type | Default | Range / options |
|---|---|---|---|
voiceA Mistral voice. Any voice id, including the account's own saved voices and Voxtral 2's presets (gb_jane_neutral), can be pasted in. | Select | voice_478 | voice_3150voice_478voice_447voice_485voice_3149voice_3147+30 |
expressivenessHow strongly the voice delivers the text, from 1 to 5. Higher is more expressive and less like the voice. Unset, Mistral picks. In the text itself, [emotion=happy] changes the emotion from that point and [pace=slow] the pace. | Number | — | 1–5, step 0.5 |
format | Select | mp3 | mp3wavflacopus |
Voices · 36
Previews aren’t available for this model’s voices.
Use this model
await mindstudio.textToSpeech({ text: "...", speechModelOverride: { model: "voxtral-tts-3", config: { voice: "voice_3150", format: "mp3", }, }, });