Voxtral Transcribe 2
Summary
Voxtral Transcribe 2 from Mistral AI introduces two speech-to-text models: Mini Transcribe V2 for batch transcription with diarization and word-level timestamps across 13 languages, and Realtime for live transcription with sub-200ms latency and open-weights. The suite includes an audio playground, edge-ready deployment options, robust privacy (GDPR/HIPAA), longer audio support, and competitive pricing, targeting enterprise use cases like meetings, contact centers, and media subtitling.