Mistral AI has introduced Voxtral 3, a language model family for real-time voice agents featuring transcription, text-to-speech, and tool integration.
Key facts
- •Mistral AI launched Voxtral 3 for real-time voice agents.
- •The stack includes Voxtral Small Transcribe 3 and Voxtral Mini TTS 3.
- •Transcription supports more than 50 languages.
- •Text-to-speech supports over 35 languages and 50 dialects with emotional adaptation.
- •Pricing for transcription starts at $0.0035 per minute.
Mistral AI has launched Voxtral 3, a new family of language models designed for real-time voice agents. The system combines listening, thinking, tool use, and speaking into a single streaming pipeline, allowing agents to interact with external systems during live conversations.
By the numbers
System Components and Capabilities
The Voxtral 3 family includes a speech-to-text model named Voxtral Small Transcribe 3, a text-to-speech model called Voxtral Mini TTS 3, and a voice agent software development kit based on LiveKit. Transcription capabilities cover more than 50 languages, while text-to-speech output supports over 35 languages and 50 dialects with emotional adaptation. During live conversations, the agents can access external systems to perform actions such as booking appointments or checking invoices.
Infrastructure and Pricing
Users can run the system on their own infrastructure or through EU-hosted application programming interfaces to maintain data control. Pricing for the transcription service starts at $0.0035 per minute.
Advertisement
This article was independently rewritten by ManyPress editorial AI from reporting originally published by The Decoder.
