Voxtral Mini Transcribe Realtime 3

Changes from previous version

Adds live streaming next to the file model Voxtral Mini Transcribe 3, at $0.008 per minute instead of $0.0035.

Release Summary

Realtime transcription model in the Voxtral 3 family, in public preview as voxtral-mini-transcribe-realtime-3-0, at $0.008 per minute. The announcement describes sub-500 ms latency across over 35 languages, with diarization, context biasing, and end-of-turn detection. Streaming delay is configurable from 80 ms to 2,400 ms. On FLEURS 35, WER is 5.7% at 2,400 ms, 6.0% at 960 ms, 6.4% at 480 ms, and 7.4% at 240 ms. The model card leaves context length blank. This entry uses the 65,536-token context and output ceiling published for Voxtral Mini Transcribe 3. Speech out is Voxtral Mini TTS 3.

Timeline

Other Voxtral versions