Oriel — AI & technology worth noticing, every day.

Voxtral 3

Web page · https://mistral.ai/news/voxtral-3

Voxtral 3: Mistral's voice agent SDK for real-time speech in 50+ languages — accurate transcription, emotional TTS, deep tool use. EU-hosted or self-deployed.

First seen by Oriel on .

AI summary

Mistral introduces Voxtral 3, a speech-model family and voice agent SDK that combine multilingual transcription, expressive speech generation, reasoning, and tool use in real-time conversations.

  • Voxtral Small Transcribe 3 supports 50 languages, diarization, context biasing, and end-of-turn detection; Voxtral Mini Transcribe Realtime 3 targets sub-500ms latency across 35+ languages.
  • Voxtral Mini TTS 3 supports 35+ languages and 50+ dialects, adapts emotion and delivery to context, and offers voice adaptation from 30+ minutes of audio.
  • The SDK, built on LiveKit, streams transcription, reasoning, tool calls, and speech without sequential stage waits; Mistral says turns complete in under one second.
  • Agents can use function calling, web search, and retrieval, and connect over SIP or WebRTC; each pipeline stage can be replaced independently.
  • The models are available through Mistral API endpoints or self-deployment, with listed pricing from $0.0035 per transcription minute and $13 per 1M TTS characters.

#speech recognition #text-to-speech #voice agents #livekit

Generated by gpt-5.6-sol. Check the original for details.

Metric trends

Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained

Points

Score reported by the source.

  • Hacker News: 765, latest observation ; +717 over observed period since

Comments

Comment count reported by the source.

  • Hacker News: 200, latest observation ; +194 over observed period since

Aliases

SourceExternal IDURL
Hacker News50168396https://mistral.ai/news/voxtral-3/ Read discussion
Mistral AI Newshttps://mistral.ai/news/voxtral-3/https://mistral.ai/news/voxtral-3/ Read discussion