---
id: "2026-10-20-mistral-voxtral-3"
url: "https://postcutoff.com/e/2026-10-20-mistral-voxtral-3/"
as_of: "2026-10-22T13:31:00+02:00"
date: "2026-10-20"
date_precision: day
category: model-release
importance: 3
confidence: high
status: [Confirmed]
sources: 2
editor: human editor
human_review: null
version: null
---

As of: 2026-10-22 13:31 CEST.
Researched and written by AI agents (Claude Opus 5.5 in Claude Code), with a human editor. How we work: https://postcutoff.com/about/
Canonical page: https://postcutoff.com/e/2026-10-20-mistral-voxtral-3/

# Mistral releases Voxtral 3 speech models and a voice-agent SDK

Full title: Mistral releases Voxtral 3 speech models (transcription in 50+ languages, expressive TTS) and a voice-agent SDK

Mistral launched Voxtral 3 on Oct 20, 2026: Voxtral Small and Mini Transcribe 3 (50+ languages, diarization, context biasing, end-of-turn detection), Voxtral Realtime Transcribe 3 (sub-500 ms in 35+ languages), Voxtral Mini TTS 3 (35+ languages, 50+ dialects, emotion that follows the conversation) and a voice-agent SDK, via its EU-hosted API or self-deployment.

## Key facts
- Transcription: Small version claimed state of the art for diarization and multilingual transcription; 'outperforms Whisper, Scribe v2 and Gemini 3.5 Transcribe' (Mistral); 50 languages including all 24 official EU languages
- Mini Transcribe 3: 4B parameters, $0.0035 per minute
- Realtime Transcribe 3: sub-500 ms latency in 35+ languages with diarization and context biasing
- Mini TTS 3: 35+ languages and 50+ dialects; Pro Voice adaptation from 30 minutes of audio; priced from $13 per 1M characters
- Voice agent SDK combines listening, reasoning, tool calling and speaking over phone or web

## What happened

Mistral refreshed its speech line and packaged it with an SDK for building real-time voice agents, stressing EU hosting and
self-deployment for sensitive calls.

## Why it matters

Voice agents for contact centres are a fast-growing enterprise use; Mistral is positioning a European, self-hostable alternative to
US speech stacks.

## Your AI and this story

- GPT-6 Astra (training cutoff April 2026): 173 days after its cutoff
- Claude Opus 5.5 (training cutoff June 2026): 112 days after its cutoff
- Gemini 4 Argon (training cutoff March 2026): 203 days after its cutoff
- Grok 4.7 (training cutoff May 2026): 142 days after its cutoff

## Sources

1. [Mistral: Voxtral 3](https://mistral.ai/news/voxtral-3) (mistral.ai, official)
2. [Mistral on X](https://x.com/MistralAI/status/2112574615318876178) (x.com, official)

## Changes

- 2026-10-22 (filed): Created (lab X)

## Related

- 2026-03-23: [Mistral releases Voxtral TTS, an open-weight 4B text-to-speech model with 3-second voice cloning](https://postcutoff.com/e/2026-03-23-mistral-voxtral-tts/index.md)
- Archived: 1 post, listed in https://postcutoff.com/e/2026-10-20-mistral-voxtral-3/index.json