best model by language
Benchmarked on the FLEURS test split at Q8_0. Error is CER — lower is better. Speed is × realtime on Handy target hardware.
accuracy vs speed
Thai · 11 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 | Qwen3-ASR-1.7Baccurate pick | 6.89 | 4.1× | 2.5× | 2.04 GB | Apache-2.0 |
| 2 | Fun-ASR-MLT-Nano-2512!overall pickfast pick | 7.99 | 8.8× | 5.5× | 0.83 GB | FunASR Model Open Source License Agreement v1.1 |
| 3 | whisper-large-v3 | 8.78 | 2.0× | 1.0× | 1.55 GB | Apache-2.0 |
| 4 | Qwen3-ASR-0.6B | 8.81 | 9.7× | 5.3× | 0.79 GB | Apache-2.0 |
| 5 | whisper-large-v2 | 12.42 | 2.0× | 1.0× | 1.55 GB | Apache-2.0 |
| 6 | whisper-large-v3-turbo | 13.15 | 3.1× | 1.4× | 0.83 GB | Apache-2.0 |
| 7 | whisper-large | 13.88 | 2.0× | 1.0× | 1.55 GB | Apache-2.0 |
| 8 | whisper-medium | 16.08 | 4.1× | 1.9× | 0.77 GB | Apache-2.0 |
| 9 | whisper-small | 22.55 | 11.7× | 5.9× | 0.25 GB | Apache-2.0 |
| 10 | whisper-base | 38.10 | 29.0× | 19.3× | 0.08 GB | Apache-2.0 |
| 11 | whisper-tiny | 54.82 | 43.1× | 39.8× | 0.04 GB | Apache-2.0 |