By Joel Khalili
Publication Date: 2026-02-04 15:32:00
Mistral AI has has released a new family of AI models that it claims will pave the way to seamless conversation between people speaking different languages.
On Wednesday, the Paris-based AI lab released two new speech-to-text models: Voxtral Mini Transcribe V2 and Voxtral Realtime. The former is designed to transcribe audio files in bulk, the latter for near real-time transcription within 200 milliseconds; Both can translate between 13 languages. Voxtral Realtime is freely available under an open source license.
With four billion parameters, the models are small enough to run locally on a phone or laptop – a first in speech recognition, according to Mistral – meaning private conversations don’t need to be moved to the cloud. Mistral says the new models are both less expensive to operate and less prone to errors than competing alternatives.
Mistral has hailed Voxtral Realtime – even though the model outputs text, not speech – as a significant step towards free-flowing conversation…
