Mistral Launches Voxtral 3 Speech Models and a Real-Time Voice Agent SDK

Mistral AI released Voxtral 3, a family of speech-to-text and text-to-speech models plus a voice agent SDK that runs listening, reasoning, tool calling and speaking as one streaming pipeline.