Points forts

Voxtral 3 is a new voice agent SDK designed for building sophisticated, real-time conversational experiences across more than 50 languages. It combines highly accurate speech transcription with a natural text-to-speech model that intelligently adapts its emotion and pacing to the context of the dialogue. The system empowers agents to listen precisely, reason through complex requests, and seamlessly integrate with external tools to execute multi-step tasks. With a strong focus on data sovereignty, Voxtral 3 can be used via an EU-hosted API or deployed directly on your own infrastructure for complete control over conversations.