Mistral launches Voxtral 3 speech models and voice agent SDK

Mistral AI on Tuesday released Voxtral 3, a family of speech-to-text and text-to-speech models alongside a voice agent SDK that combines listening, reasoning, tool calling and speaking in one real-time streaming pipeline. The models are available today through the Mistral API and can also be self-deployed on a company's own infrastructure.

On transcription, Voxtral Small Transcribe 3 supports 50 languages, including the 24 official EU languages, with speaker diarization, context biasing for domain-specific terms and end-of-turn detection. Mistral says it achieves state-of-the-art results in diarization and multilingual transcription and outperforms Whisper, Scribe v2 and Gemini 3.5 Transcribe. It costs $0.0040 per minute. A smaller 4B-parameter Voxtral Mini Transcribe 3 costs $0.0035 per minute in batch, and Voxtral Mini Transcribe Realtime 3, built for agents, costs $0.008 per minute with what Mistral says is sub-500ms latency.

For speech output, Voxtral Speech 3 supports 35+ languages and 50+ dialects, adjusting emotion, pacing and emphasis to the conversation, and costs $13 per 1 million characters. Mistral says that undercuts comparable models from OpenAI, Deepgram, ElevenLabs Flash and Gemini Flash. A Pro Voice feature adapts a speaker's voice and style across languages from 30 minutes of audio.

The voice agent SDK is built on LiveKit and chains a realtime transcription model, a general model such as Mistral Small for reasoning and tool calls, and Voxtral Speech 3 for replies. Mistral says each turn completes in under a second. Agents can handle calls over SIP, run in browsers and apps over WebRTC, and connect to existing telephony and contact-center software.