Voice mode in le Chat is powered by a specialized speech-to-text (STT) model developed by Mistral AI: voxtral-mini-2507.
This model is optimized for fast, accurate transcription across a range of contexts and languages, including:
English
Français (French)
Deutsch (German)
Español (Spanish)
Italiano (Italian)
Nederlands (Dutch)
Português (Portuguese)
हिन्दी (Hindi)
Using our own model helps provide a high-quality, seamless voice experience in le Chat.
🔎 For details on capabilities and design choices, read the official announcement.