Voxtral 3

Voxtral 3 is a voice agent SDK and a family of speech models designed for real-time, multi-step conversations in over 50 languages. It combines accurate speech-to-text transcription, natural text-to-speech with emotional adaptability, and deep tool integration for automating complex workflows. The system is designed for applications like contact centers and virtual assistants, with transcription models that outperform competitors in multilingual benchmarks. Available via an EU-hosted API or for self-deployment, it provides control over data sovereignty and includes a "Pro Voice" feature for high-fidelity voice adaptation.