Quantization format
IQ4_NL
IQ4_NL has a nominal rate of — bits per weight. We have no measured files carrying this label yet.
From the file· 840 files measured
Nominal bpw
—
from the block layout
Measured average
4.779
840 files
Range
2.217–32.274
varies by architecture
File sizes
0.03 GiB+
up to 539.67 GiB
Real files
one per model, most downloaded first
| Model | Size● | Effective bpw● | vs nominal● |
|---|---|---|---|
| Qwen3.6-27B | 14.97 GiB | 4.628 | — |
| Qwen3.6-35B-A3B | 19.33 GiB | 4.618 | — |
| Qwen3.5-9B | 5.00 GiB | 4.451 | — |
| gemma-4-26B-A4B-it | 13.69 GiB | 4.430 | — |
| gemma-4-12B-it | 6.26 GiB | 4.493 | — |
| Qwen3.5-4B | 2.40 GiB | 4.429 | — |
| Hy3 | 158.14 GiB | 4.546 | — |
| gemma-4-E4B-it | 4.50 GiB | 4.838 | — |
| Qwen3-Coder-30B-A3B-Instruct | 16.12 GiB | 4.536 | — |
| gemma-4-31B-it | 16.10 GiB | 4.422 | — |
| Qwen3-VL-30B-A3B-Instruct | 16.12 GiB | 4.457 | — |
| Llama-3.2-1B-Instruct | 0.72 GiB | 5.004 | — |
| gemma-4-E2B-it | 2.83 GiB | 4.749 | — |
| Qwen3-8B | 4.46 GiB | 4.682 | — |
| Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-MTP | 32.65 GiB | 10.095 | — |
| Qwen3.5-0.8B | 0.47 GiB | 4.642 | — |
| Qwen3-4B | 2.22 GiB | 4.736 | — |
| Llama-3.1-8B-Instruct | 4.36 GiB | 4.660 | — |
| Ornith-1.0-35B | 18.50 GiB | 4.584 | — |
| Llama-3.2-3B-Instruct | 1.79 GiB | 4.774 | — |
| ThinkingCap-Qwen3.6-27B | 15.20 GiB | 4.774 | — |
| Qwythos-9B-v2 | 5.23 GiB | 4.654 | — |
| Qwen3.5-122B-A10B | 67.26 GiB | 4.619 | — |
| Qwen3-Coder-Next | 42.03 GiB | 4.532 | — |
| gemma-3-1b-it | 0.67 GiB | 5.776 | — |
●From the filewhat these mean