Join waitlist
Back to products
Voxtral

Voxtral

Frontier open source speech understanding models

Website mistral.ai
Overview

What it is

- We’re committed to empower the AI community with open technology. Our open models sets the bar for efficiency, and are available for free, with fully permissive license. - Our optimized commercial models are designed for performance and are available via our flexible deployment options.

Intent

I need it when

Add voice cloning features to applications without building custom models

Voxtral includes voice cloning functionality as part of its TTS API, allowing users to generate speech in custom voices without requiring separate model training or infrastructure

Generate natural-sounding speech from text for applications, content, or accessibility

Voxtral provides state-of-the-art text-to-speech generation at $0.016 per 1,000 characters, enabling developers and content creators to convert written content into high-quality audio output with voice cloning capabilities

Create accessible content by converting documents and reports to audio format

Voxtral enables bulk text-to-speech conversion for accessibility, allowing organizations to automatically generate audio versions of written materials for users who prefer or require audio content

Integrate audio generation into production applications with enterprise-grade reliability

Voxtral is available through Mistral's Enterprise APIs with regional data processing controls, system-level SLAs, increased rate limits, and premium support for mission-critical deployments

Drop

Not a fit when

  • User needs real-time voice conversation with sub-100ms latency; Voxtral is a text-to-speech API, not a real-time conversational voice system
  • User requires on-device processing without external API calls; Voxtral is a cloud-based API requiring internet connectivity
  • User needs support for rare or constructed languages; Voxtral supports standard natural languages only
  • User has strict data residency requirements outside EU; Mistral servers are hosted in the EU
  • User needs unlimited voice generation at fixed monthly cost; Voxtral charges per 1,000 characters with no monthly cap option
Commercials

Pricing

Pay-per-use API pricing based on character volume View pricing