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

Dutch · 21 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 Voxtral-Small-24B-2507 5.12 — — 24.04 GB Apache-2.0
2 whisper-large-v3 5.42 2.0× 1.0× 1.55 GB Apache-2.0
3 whisper-large-v3-turboaccurate pick 5.98 3.1× 1.4× 0.83 GB Apache-2.0
4 Voxtral-Mini-4B-Realtime-2602 6.09 0.8× 0.6× 4.41 GB Apache-2.0
5 canary-1b-v2!overall pickfast pick 6.28 14.1× 8.3× 1.07 GB CC-BY-4.0
6 Voxtral-Mini-3B-2507 6.57 1.1× 0.6× 4.66 GB Apache-2.0
7 parakeet-ultra 6.70 23.0× 13.3× 0.69 GB CC-BY-4.0
8 whisper-large-v2 6.76 2.0× 1.0× 1.55 GB Apache-2.0
9 orukeet 6.93 25.1× 13.6× 0.69 GB CC-BY-SA-4.0
10 cohere-transcribe-03-2026! 7.16 8.1× 4.3× 2.25 GB Apache-2.0
11 Qwen3-ASR-1.7B 7.43 4.1× 2.5× 2.04 GB Apache-2.0
12 parakeet-tdt-0.6b-v3 7.66 25.1× 13.7× 0.69 GB CC-BY-4.0
13 whisper-large 8.48 2.0× 1.0× 1.55 GB Apache-2.0
14 parakeet-primeline 8.49 25.0× 13.6× 0.69 GB CC-BY-4.0
15 whisper-medium 10.40 4.1× 1.9× 0.77 GB Apache-2.0
16 nemotron-3.5-asr-streaming-0.6b 13.61 17.1× 10.6× 0.70 GB OpenMDW-1.1
17 Qwen3-ASR-0.6B 13.90 9.7× 5.3× 0.79 GB Apache-2.0
18 whisper-small 18.48 11.7× 5.9× 0.25 GB Apache-2.0
19 whisper-base 36.75 29.0× 19.3× 0.08 GB Apache-2.0
20 Fun-ASR-MLT-Nano-2512! 42.97 8.8× 5.5× 0.83 GB FunASR Model Open Source License Agreement v1.1
21 whisper-tiny 54.86 43.1× 39.8× 0.04 GB Apache-2.0