Mistral Releases Voxtral 3 Voice Model Family and Voice Agent SDK
Understand the story
On October 20, 2026, Mistral AI officially released the Voxtral 3 voice model family, along with a voice agent SDK built on LiveKit. The Decoder subsequently reported more details: the Voxtral 3 family includes the speech-to-text model Voxtral Small Transcribe 3 and the text-to-speech model Voxtral Mini TTS 3, plus a LiveKit-based voice agent SDK. According to the report, the solution integrates listening, thinking, tool calling, and speaking into a single streaming pipeline, targeting real-time voice agent scenarios. This is the latest update to Mistral's voice product line.
AI overview of reporting · Updated 1 h前
Reporting timeline
Explore different aspects of the story through its reporting.
- The DecoderMistral AI launches Voxtral 3, a full voice agent stack that listens, thinks, and talks in real time
Mistral AI released the Voxtral 3 model family for real-time voice agents, including the speech-to-text model Voxtral Small Transcribe 3, the text-to-speech model Voxtral Mini TTS 3, and a LiveKit-based voice agent SDK, integrating listening, thinking, tool calling, and speaking into a single streaming pipeline.
- Mistral AISelectedVoxtral 3
Mistral released Voxtral 3, a family of speech transcription and speech synthesis models plus a LiveKit-based SDK for voice agents.
Story heat trend
Current heat18·Comparable peak18（10/21 02:00）·Comparable change over 24 hours –
The trend compares the same continuously observed sources, which may cover fewer sources than the current heat score. Hover or select the chart to inspect each hour; use the left and right arrow keys to navigate.