Mistral releases Voxtral 3 speech models and voice agent SDK, ranked by Artificial Analysis
Article timeline
The articles in this story. Times are ET.
OpenRouterOfficialMistral releases Voxtral 3 speech models on OpenRouterAIMistral has released two speech-to-text models and one text-to-speech model under the Voxtral 3 name, covering more than 50 languages. OpenRouter says Voxtral 3 is now available on its platform.
- The DecoderNewsPickMistral AI launches Voxtral 3, a voice agent stack for real-time listening and speaking
AIMistral AI has launched Voxtral 3, a family of models for real-time voice agents, including a speech-to-text model, a text-to-speech model, and a LiveKit-based agent SDK. The system can run on user infrastructure or through EU-hosted APIs, with transcription starting at $0.0035 per minute. Agents can call external systems mid-conversation, such as booking appointments or checking invoices.
Sophia YangOfficialMistral releases Voxtral 3 transcription and speech models with a voice agent SDKAIMistral announces Voxtral 3, a set of Transcribe and Speech models that power its new voice agents API. The release also includes a voice agent SDK for building voice applications.
Guillaume Lample @ NeurIPS 2024OfficialMistral releases Voxtral 3 speech models and voice agent SDKAIMistral introduces Voxtral 3, a family of speech models including Mini and Small ASR models for batch and real-time streaming transcription. It also adds a TTS model for expressive multilingual speech and a voice agent SDK supporting low-latency conversations and tool calling. The models are available through an API or on users' own infrastructure.
DesignArenaOfficialMistral's Voxtral TTS 3 comes to Design Arena for real-time voice agentsAIDesign Arena says Mistral's Voxtral TTS 3 is now available on its platform. The model supports 35+ languages, natural turn-taking, and sub-second response times for real-time voice agents. Its streaming pipeline combines speech recognition, reasoning, tool calling, and speech generation for tasks such as billing inquiries and appointment booking.
Arthur MenschOfficialPickMistral releases a suite of state-of-the-art voice models called Voxtral 3AIArthur Mensch, co-founder of Mistral, announces a comprehensive suite of state-of-the-art voice models and links to Mistral's announcement, Voxtral 3. He says voice will become the most important interface to AI systems, and that Mistral deploys voice models physically close to its reasoning models.
Artificial AnalysisOfficialMistral releases Voxtral 3 speech models, Small Transcribe 3 ranks #3 on speech-to-textAIArtificial Analysis evaluated Mistral's Voxtral 3 family, which includes a text-to-speech model and three speech-to-text models. Voxtral Small Transcribe 3 ranks #3 on the Speech to Text leaderboard with a 2.1% AA-WER Index, ahead of ElevenLabs Scribe v2 at 2.2%. Voxtral 3 TTS is priced at $13 per 1M characters, below Gemini 3.8 Flash TTS at $16.5 and Eleven v4 Turbo at $40.
- Artificial AnalysisOfficialMistral releases Voxtral 3 speech models, Small Transcribe 3 ranks #3 on AA leaderboard
AIMistral has released Voxtral 3, a text-to-speech model and three speech-to-text models, which Artificial Analysis evaluated. Voxtral Small Transcribe 3 ranks #3 on the Speech to Text leaderboard with a 2.1% AA-WER Index, ahead of ElevenLabs Scribe v2 at 2.2%. Voxtral 3 TTS is priced at $13 per 1M characters, below Gemini 3.8 Flash TTS at $16.5 and Eleven v4 Turbo at $40.
Mistral AIOfficialMistral AI introduces Voxtral 3 transcription and speech models for voice agentsAIMistral AI introduces Voxtral 3, which it says are state-of-the-art Transcribe and Speech models. The models power the company's new voice agents API.
- Mistral AIOfficialPickMistral releases Voxtral 3 speech models and a voice agent SDK
AIMistral AI releases Voxtral 3, a family of speech-to-text and text-to-speech models plus a voice agent SDK that runs listening, reasoning, tool calling, and speaking in one real-time pipeline. The speech-to-text models support over 50 languages, and Voxtral Speech 3 is priced at $13 per 1M characters. The stack is available through Mistral's EU-hosted API or for self-deployment on customers' own infrastructure.