Quantization format
Q4_1
Q4_1 has a nominal rate of — bits per weight. We have no measured files carrying this label yet.
From the file· 878 files measured
Nominal bpw
—
from the block layout
Measured average
5.371
878 files
Range
1.431–30.248
varies by architecture
File sizes
0.02 GiB+
up to 598.99 GiB
Real files
one per model, most downloaded first
| Model | Size● | Effective bpw● | vs nominal● |
|---|---|---|---|
| Qwen3.6-27B | 16.07 GiB | 4.968 | — |
| Qwen3.6-35B-A3B | 21.30 GiB | 5.089 | — |
| Qwen3.5-9B | 5.44 GiB | 4.838 | — |
| gemma-4-26B-A4B-it | 15.04 GiB | 4.866 | — |
| gemma-4-12B-it | 6.89 GiB | 4.948 | — |
| Qwen3.5-4B | 2.59 GiB | 4.780 | — |
| Hy3 | 174.90 GiB | 5.028 | — |
| gemma-4-E4B-it | 4.73 GiB | 5.077 | — |
| Qwen3-Coder-30B-A3B-Instruct | 17.87 GiB | 5.029 | — |
| gemma-4-31B-it | 17.81 GiB | 4.891 | — |
| Qwen3-VL-30B-A3B-Instruct | 17.87 GiB | 4.942 | — |
| FLUX.2-klein-9B | 5.74 GiB | 5.429 | — |
| LTX-2.3 | 12.81 GiB | 11.967 | — |
| Llama-3.2-1B-Instruct | 0.77 GiB | 5.384 | — |
| gemma-4-E2B-it | 2.94 GiB | 4.926 | — |
| gpt-oss-20b | 10.78 GiB | 4.306 | — |
| Qwen3-8B | 4.89 GiB | 5.126 | — |
| Qwen3.5-0.8B | 0.50 GiB | 4.902 | — |
| Qwen3-4B | 2.42 GiB | 5.164 | — |
| Llama-3.1-8B-Instruct | 4.78 GiB | 5.111 | — |
| Ornith-1.0-35B | 20.46 GiB | 5.072 | — |
| Llama-3.2-3B-Instruct | 1.95 GiB | 5.213 | — |
| ThinkingCap-Qwen3.6-27B | 16.60 GiB | 5.213 | — |
| whisper-medium | 0.45 GiB | 5.050 | — |
| Wan2.2-I2V-A14B | 8.62 GiB | 5.184 | — |
●From the filewhat these mean