Mistral AI launches Voxtral 3 for real-time voice agents
Mistral AI launches Voxtral 3, a full voice agent stack that listens, thinks, and talks in real time
- Who
- Mistral
- What
- introduces the Voxtral 3 family of speech models and voice agent SDK
- Quote
Mistral AI has launched Voxtral 3, a new family of language models built for real-time voice agents.
What happened
Mistral AI has launched Voxtral 3, a model family for real-time voice agents that combines Voxtral Small Transcribe 3, Voxtral Mini TTS 3, and a LiveKit-based voice agent SDK. Transcription covers more than 50 languages, while speech output supports over 35 languages and 50 dialects with emotional adaptation. Mistral says the system supports tool use during conversations and deployment on users’ infrastructure or through EU-hosted APIs, with transcription pricing starting at $0.0035 per minute.
The same story is shown in the picks as 《Mistral AI releases Voxtral 3 speech models and a voice agent SDK》
Reported by 4 sources
The story so far
Mistral AI released Voxtral 3 transcription and speech generation models, a voice agents API and a LiveKit-based voice agent SDK, with the models available through its EU-hosted API or for deployment on customers’ infrastructure. Organizations seeking to run the full stack on-premises or in a private cloud must contact Mistral’s team. Transcription supports more than 50 languages, while speech generation supports more than 35 languages and more than 50 dialects. The SDK combines listening, reasoning, tool calling and speaking in a streaming pipeline that Mistral says completes conversational turns in under one second. The Decoder reported that transcription pricing starts at $0.0035 per minute. Artificial Analysis ranked Voxtral Small Transcribe 3 third on its Speech to Text leaderboard with a 2.1% AA-WER Index. It reported final-transcript AA-WER of 3.3% at 960 ms and 3.6% at 480 ms for Mini Transcribe Realtime 3, compared with its predecessor’s 5.2%. Artificial Analysis also ranked Voxtral 3 TTS 12th in both its Provider Voice and Controlled Voice arenas and reported a price of $13 per million characters.
Keep reading
Latest daily · Issue 27TorchNeuron brings native PyTorch support for Hugging Face models on AWS TrainiumTrending now
Tomorrow at 08:00, a briefing like this arrives in your inbox.
The same kind of briefing as the one above: a few stories worth reading. Free. Mail starts after you confirm, and every letter has a one-click unsubscribe. Read a past issue · 2026-10-20