Mistral Launches Voxtral 3 Speech Models and Voice Agents API

Key Info

Mistral AI announced Voxtral 3, comprising state-of-the-art Transcribe and Speech models that power Mistral's new voice agents API. The release also includes a voice agent SDK built on LiveKit.

Highlights

• Voxtral 3 includes Transcribe and Speech models described as state-of-the-art.
• The models power Mistral's new voice agents API.
• A voice agent SDK is introduced, built on LiveKit.
• The pipeline combines Transcribe (listening), a reasoning model (thinking and tool calls), and Speech (replying), streamed as one pipeline for natural conversation.