With 4.6B active parameters (26.6B total), SOV-Transcribe-1 sits on the open weights frontier of accuracy vs. model size. It is more accurate than Voxtral Small with ~5x fewer active parameters (4.6B vs. 24B), and than Nemotron 3.5 Transcribe at a similar active parameter count (4.6B vs. 4B)
SOV-Transcribe-1 beats larger transcription models on accuracy with 4.6B active parameters
AIArtificial Analysis says SOV-Transcribe-1, with 4.6B active parameters (26.6B total), sits on the open weights frontier of accuracy versus model size. It is more accurate than Voxtral Small, which has 24B active parameters, and more accurate than Nemotron 3.5 Transcribe, which has 4B active parameters.
Source details
Artificial Analysis · x.comPublished
Original title: With 4.6B active parameters (26.6B total), SOV-Transcribe-1 sits on the open weights frontier of accuracy vs. model size. It is more accu...