AI Development

Mistral AI launches Voxtral 3 speech models and voice agent SDK

October 20, 2026 · Medium importance · product
Editorial Analysis

Mistral released a full speech stack including multilingual transcription and TTS models plus a LiveKit-based voice agent SDK for real-time listening, reasoning, tool use and speaking. The models support 50+ languages for transcription and are available via API or self-deployment.

At a Glance
Date
October 20, 2026
Importance
Medium 3/5
Category
product
Axis of Change
Capability Gain
Organizations
Mistral AI
Models Affected
Voxtral 3 Voxtral Small Transcribe 3 Voxtral Speech 3
Sources
  1. 01 mistral.ai
  2. 02 docs.mistral.ai
  3. 03 docs.mistral.ai
Related Developments
Oct 14, 2026 · product

xAI releases Grok Voice 3.0 speech-to-speech model topping independent Speech-to-Speech Index
Oct 12, 2026 · product

Forbes details emerging category of specialized AI decision models beyond Jev
Oct 8, 2026 · product

OpenAI expands GPT-6 with Intelligent UI to free ChatGPT tiers