Mistral Voxtral TTS 3
Mistral's text-to-speech model with preset voices in eight languages; emotion and pace tags written into the text, such as [emotion=happy] or [pace=slow], direct the delivery. In public preview.
Voxtral TTS 3 is a cloud text-to-speech model built by Mistral. It converts written text into natural, spoken audio. It offers a choice of 8 voices. It runs through OpenRouter using your own API key, from $0.013 per 1,000 characters.
- Released
- Oct 2026
- Pricing
- $0.013 / 1k chars
- Type
- Text-to-speech
- Voices
- 8 to choose from
Samples
Generated with Voxtral TTS 3 via OpenRouter: the same three scripts read by every text-to-speech model in the catalog, so the only thing that changes between two models’ clips is the model. Each model uses its own default voice; there is no shared voice to hold constant.
Baseline naturalness and pacing
The last train had already left, but she decided to walk anyway. The city was quieter than she remembered, and for the first time in weeks, she wasn't in a hurry.
Emotion, emphasis, and pauses
Wait — you're telling me it actually worked? After all that? I can't believe it. Honestly, I thought we'd lost the whole thing.
Decimals, percentages, version numbers
The API returned 3,481 results in 0.42 seconds, a 12 percent improvement over version 2.5. Latency at the 99th percentile dropped from 840 milliseconds to 610.
Mistral
Mistral is a Paris-based lab that ships fast, efficient open-weight language models with strong reasoning per parameter.
mistral.ai ↗Mistral's largest open-weight model, a multimodal mixture-of-experts built for reasoning, coding and agent work, with vision, tool calling and a 512K-token context window.
Mistral's flagship transcription model, which returns a timestamped transcript and detects the language on its own. In public preview.
Mistral's compact transcription model, a lower-cost sibling of Voxtral Small Transcribe 3 with timestamps and language detection. In public preview.
Mistral's 24B instruction-tuned model with vision input, robust function calling, and a 128K context.
Mistral's 14B dense edge model (Apache 2.0) tuned for fast on-device reasoning and tool use.
Mistral's efficient 8B dense edge model (Apache 2.0) for on-device chat, reasoning, and tool use.