Quantization format
UD_IQ1_M
UD_IQ1_M has a nominal rate of — bits per weight. We have no measured files carrying this label yet.
From the file· 178 files measured
Nominal bpw
—
from the block layout
Measured average
2.311
178 files
Range
1.467–4.479
varies by architecture
File sizes
0.21 GiB+
up to 288.02 GiB
Real files
one per model, most downloaded first
| Model | Size● | Effective bpw● | vs nominal● |
|---|---|---|---|
| Qwen3.6-35B-A3B | 9.36 GiB | 2.236 | — |
| DeepSeek-V4-Flash | 80.93 GiB | 2.389 | — |
| Qwen3-Coder-30B-A3B-Instruct | 8.97 GiB | 2.523 | — |
| Qwen3-VL-30B-A3B-Instruct | 9.00 GiB | 2.489 | — |
| Llama-3.2-1B-Instruct | 0.41 GiB | 2.843 | — |
| Qwen3-8B | 2.23 GiB | 2.341 | — |
| Qwen3-4B | 1.06 GiB | 2.274 | — |
| GLM-5.2 | 212.80 GiB | 2.426 | — |
| Llama-3.1-8B-Instruct | 2.13 GiB | 2.284 | — |
| Ornith-1.0-35B | 10.29 GiB | 2.549 | — |
| Llama-3.2-3B-Instruct | 0.89 GiB | 2.392 | — |
| Kimi-K2.7-Code | 283.04 GiB | 2.297 | — |
| Qwen3.5-122B-A10B | 31.87 GiB | 2.189 | — |
| Qwen3-Coder-Next | 20.21 GiB | 2.179 | — |
| gemma-3-1b-it | 0.52 GiB | 4.479 | — |
| Qwen3-14B | 3.79 GiB | 2.202 | — |
| Qwen3-VL-4B-Instruct | 1.06 GiB | 2.061 | — |
| Qwen3-0.6B | 0.21 GiB | 2.350 | — |
| Qwen3-VL-2B-Instruct | 0.52 GiB | 2.113 | — |
| Qwen2.5-VL-7B-Instruct | 2.05 GiB | 2.123 | — |
| Qwen3-30B-A3B | 9.00 GiB | 2.533 | — |
| Qwen3-1.7B | 0.52 GiB | 2.213 | — |
| Laguna-S-2.1 | 33.19 GiB | 2.425 | — |
| Qwen3-4B-Instruct-2507 | 1.06 GiB | 2.274 | — |
| gemma-3-12b-it | 3.03 GiB | 2.133 | — |
●From the filewhat these mean