Quantization format
Q4_K_L
Q4_K_L has a nominal rate of — bits per weight. We have no measured files carrying this label yet.
From the file· 565 files measured
Nominal bpw
—
from the block layout
Measured average
5.203
565 files
Range
3.776–8.749
varies by architecture
File sizes
0.10 GiB+
up to 541.31 GiB
Real files
one per model, most downloaded first
| Model | Size● | Effective bpw● | vs nominal● |
|---|---|---|---|
| Qwen3.6-35B-A3B | 21.11 GiB | 5.043 | — |
| gemma-4-26B-A4B-it | 16.03 GiB | 5.188 | — |
| gemma-4-12B-it | 7.36 GiB | 5.289 | — |
| Qwen3.5-4B | 2.95 GiB | 5.437 | — |
| Hy3 | 169.99 GiB | 4.887 | — |
| gemma-4-31B-it | 18.57 GiB | 5.101 | — |
| Llama-3.2-1B-Instruct | 0.81 GiB | 5.640 | — |
| gemma-4-E2B-it | 3.85 GiB | 6.448 | — |
| Qwen3-8B | 5.11 GiB | 5.362 | — |
| Qwen3.5-0.8B | 0.60 GiB | 5.873 | — |
| Llama-3.1-8B-Instruct | 4.95 GiB | 5.291 | — |
| Ornith-1.0-35B | 20.27 GiB | 5.024 | — |
| Llama-3.2-3B-Instruct | 1.97 GiB | 5.266 | — |
| ThinkingCap-Qwen3.6-27B | 17.43 GiB | 5.473 | — |
| Qwythos-9B-v2 | 6.34 GiB | 5.638 | — |
| Qwen3.5-122B-A10B | 72.81 GiB | 5.000 | — |
| Qwen3-Coder-Next | 45.60 GiB | 4.916 | — |
| Qwen2.5-32B-Instruct | 19.03 GiB | 4.988 | — |
| Qwen3.5-35B-A3B | 21.11 GiB | 5.043 | — |
| Qwen2.5-Coder-7B-Instruct | 4.74 GiB | 5.344 | — |
| Qwen2.5-7B-Instruct | 4.74 GiB | 5.344 | — |
| Qwen3-0.6B | 0.56 GiB | 6.383 | — |
| Qwen3-VL-2B-Instruct | 1.10 GiB | 4.447 | — |
| Jan-v3-4B-base-instruct | 2.80 GiB | 5.449 | — |
| Qwen2.5-1.5B-Instruct | 0.97 GiB | 5.403 | — |
●From the filewhat these mean