handy models

speech-to-text benchmarks for transcribe.cpp, the engine behind Handy

best model by language

Benchmarked on the FLEURS test split at Q8_0. Error is WER — lower is better. Speed is × realtime on Handy target hardware.

accuracy vs speed

Czech · 17 models

AMD Ryzen 4750U GPU (Vulkan)
pareto frontier (no model is both faster and more accurate) other models
highlighted rows are recommended picks or on pareto frontier
1 canary-1b-v2!overall pickaccurate pickfast pick 8.56 14.1× 8.3× 1.07 GB CC-BY-4.0
2 orukeet 10.22 25.1× 13.6× 0.69 GB CC-BY-SA-4.0
3 whisper-large-v3 10.50 2.0× 1.0× 1.55 GB Apache-2.0
4 parakeet-ultra 11.04 23.0× 13.3× 0.69 GB CC-BY-4.0
5 whisper-large-v3-turbo 11.81 3.1× 1.4× 0.83 GB Apache-2.0
6 parakeet-tdt-0.6b-v3 12.31 25.1× 13.7× 0.69 GB CC-BY-4.0
7 whisper-large-v2 14.42 2.0× 1.0× 1.55 GB Apache-2.0
8 parakeet-primeline 15.05 25.0× 13.6× 0.69 GB CC-BY-4.0
9 whisper-large 18.77 2.0× 1.0× 1.55 GB Apache-2.0
10 whisper-medium 22.85 4.1× 1.9× 0.77 GB Apache-2.0
11 Qwen3-ASR-1.7B 22.97 4.1× 2.5× 2.04 GB Apache-2.0
12 nemotron-3.5-asr-streaming-0.6b 23.00 17.1× 10.6× 0.70 GB OpenMDW-1.1
13 whisper-small 40.57 11.7× 5.9× 0.25 GB Apache-2.0
14 Qwen3-ASR-0.6B 44.50 9.7× 5.3× 0.79 GB Apache-2.0
15 Fun-ASR-MLT-Nano-2512! 53.56 8.8× 5.5× 0.83 GB FunASR Model Open Source License Agreement v1.1
16 whisper-base 70.14 29.0× 19.3× 0.08 GB Apache-2.0
17 whisper-tiny 87.17 43.1× 39.8× 0.04 GB Apache-2.0