Voxtral
Voxtral is Mistral AI's family of speech models for transcription, realtime speech recognition and text-to-speech. Voxtral 3 adds a voice agent SDK and covers 50+ languages for transcription.
About Voxtral
What is Voxtral?
Voxtral is Mistral AI's family of speech models for transcription, realtime speech recognition and text-to-speech, with a voice agent SDK added in Voxtral 3.
Voxtral 3 models
- Small Transcribe 3: highest-accuracy batch transcription with diarization and context biasing ($0.004/min).
- Mini Transcribe 3: low-latency batch transcription ($0.0035/min).
- Mini Transcribe Realtime 3: streaming, sub-500 ms latency, end-of-turn detection ($0.008/min).
- Mini TTS 3: emotion and pacing control, voice cloning ($13 per 1M characters).
Voice agent SDK
Built on LiveKit: listen, reason with a Mistral LLM, call tools and speak in one streaming pipeline, over SIP or WebRTC.
Languages and deployment
50+ languages for transcription, 35+ for TTS. EU-hosted API with US and global endpoints, self-deployment and on-premises options. The original 2025 Voxtral models are open weights under Apache 2.0.
Company
Mistral AI, Paris, founded 2023.
Value & Audience
Value Proposition
One vendor for transcription, realtime recognition, text-to-speech and a voice agent pipeline, at per-minute prices and with EU hosting and self-deployment options.
Problem Solved
Building voice products usually means stitching together separate speech-to-text, LLM and TTS providers with different latency, language coverage and data-residency terms.
Tech Stack
Other
Social & Links
Loading ratings...
Loading comments...
Gallery
Feedback & Roadmap
Voxtral FAQ
Common questions about Voxtral's features, pricing, and use cases
What does Voxtral do?
Mistral's speech-to-text, TTS and voice agent models One vendor for transcription, realtime recognition, text-to-speech and a voice agent pipeline, at per-minute prices and with EU hosting and self-deployment options.
Is Voxtral free to use?
Voxtral is a paid product, starting at Transcription - $0.004/min (Small Transcribe 3), $0.0035/min (Mini Transcribe 3). Full pricing details are listed on the Voxtral product page on TechLogHub.
What problem does Voxtral solve?
Building voice products usually means stitching together separate speech-to-text, LLM and TTS providers with different latency, language coverage and data-residency terms.
Who is Voxtral built for?
Voxtral is built for Developers, Data Scientists & ML Engineers, Startups, Enterprises, Customer Support Teams.
What tech stack does Voxtral use?
Voxtral is built with LiveKit.
What category is Voxtral listed under?
Voxtral is listed in the AI Tools & Platforms category on TechLogHub, alongside curated alternatives with community ratings and pricing comparisons.
Voxtral appears in
You May Also Like
View all alternatives to Voxtral
Jev
Jev by TypeSafe AI is a model API that takes state plus typed questions and returns Choice, Score and true/false answers with calibrated probabilities instead of generated text.
View ProductRicebowl AI
Ricebowl AI is an independent commercial AI creation platform for image and video workflows. Generate watermark-free, commercial-ready exports with flagship models including Sora 2, Veo, Kling, Flux and more for marketing teams, ads, and product creatives.
View Product
ToSong
ToSong is an AI song maker: paste lyrics or describe a song and get melody, arrangement and vocals in one generation. 20 free credits on signup.
View Product
Quick I Ching
Use the I Ching online for free with Three-Coin, Yarrow Stalk, Mei Hua Yi Shu, or Manual Cast. Get your hexagram, changing lines, relating hexagram when applicable, and a clear I Ching reading.
View ProductOne email a week
TechLogHub tracks 624 tools and 541 open-source projects. Get the week's new launches, release roundups and open-source picks — nothing else.
Get the TechLogHub digest
One email a week: new and trending developer tools, fresh comparisons, and what shipped. Unsubscribe in one click.