Quantization format
IQ2_M
IQ2_M has a nominal rate of — bits per weight. We have no measured files carrying this label yet.
From the file· 745 files measured
Nominal bpw
—
from the block layout
Measured average
2.988
745 files
Range
1.178–10.009
varies by architecture
File sizes
0.02 GiB+
up to 321.02 GiB
Real files
one per model, most downloaded first
| Model | Size● | Effective bpw● | vs nominal● |
|---|---|---|---|
| Qwen3.6-35B-A3B | 12.07 GiB | 2.885 | — |
| gemma-4-26B-A4B-it | 9.97 GiB | 3.225 | — |
| gemma-4-12B-it | 4.60 GiB | 3.305 | — |
| Qwen3.5-4B | 1.82 GiB | 3.354 | — |
| Hy3 | 93.14 GiB | 2.678 | — |
| gemma-4-31B-it | 11.78 GiB | 3.236 | — |
| gemma-4-E2B-it | 2.38 GiB | 3.987 | — |
| Qwen3-8B | 2.84 GiB | 2.981 | — |
| Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-MTP | 22.16 GiB | 6.853 | — |
| Qwen3.5-0.8B | 0.37 GiB | 3.651 | — |
| Llama-3.1-8B-Instruct | 2.75 GiB | 2.937 | — |
| Ornith-1.0-35B | 11.24 GiB | 2.785 | — |
| Qwopus3.6-27B-Coder | 9.74 GiB | 3.011 | — |
| ThinkingCap-Qwen3.6-27B | 10.13 GiB | 3.180 | — |
| Qwythos-9B-v2 | 3.64 GiB | 3.237 | — |
| Qwen3.5-122B-A10B | 41.39 GiB | 2.842 | — |
| Qwen3-Coder-Next | 24.31 GiB | 2.620 | — |
| Qwen2.5-32B-Instruct | 10.49 GiB | 2.751 | — |
| Qwen3.5-35B-A3B | 12.07 GiB | 2.885 | — |
| Qwen2.5-Coder-7B-Instruct | 2.59 GiB | 2.921 | — |
| Qwen2.5-7B-Instruct | 2.59 GiB | 2.921 | — |
| Qwen3-0.6B | 0.31 GiB | 3.531 | — |
| Qwen3-VL-2B-Instruct | 0.65 GiB | 2.614 | — |
| Jan-v3-4B-base-instruct | 1.56 GiB | 3.047 | — |
| Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking | 13.76 GiB | 2.990 | — |
●From the filewhat these mean