Mistral releases Voxtral Transcribe 2 with 13-language real-time ASR under 500 ms latency
Mistral AI released Voxtral Transcribe 2 on August 14, combining batch and real-time ASR across 13 languages. Voxtral Mini 4B real-time model achieves under 500 ms latency. Voxtral Mini Transcribe V2 is priced at $0.003/minute.
Mistral AI released Voxtral Transcribe 2 on August 14, combining batch diarization and real-time automatic speech recognition (ASR), designed for multilingual production workloads. The system supports 13 languages including English, Spanish, Chinese, and Hindi.
The Voxtral Mini 4B real-time model achieves under 500 ms latency, with accuracy comparable to offline systems. Voxtral Mini Transcribe V2 focuses on batch transcription with speaker diarization, priced at $0.003 per minute.
Voxtral Transcribe 2 delivers significantly better non-English performance than competitors. Mistral is positioning it as a leading multilingual ASR solution. Mistral also launched Voxtral TTS (3B-parameter open-source model, claiming a 63% win rate against ElevenLabs Flash v2.5) and cross-lingual voice cloning.
In parallel, Mistral announced enterprise strategy: a five-year cloud agreement with European carrier KLM deploying Gemini AI tools and DeepMind models to 35,000 employees; a target of 1GW of European compute by 2030; and interoperability with Zhipu GLM-5.2.