Quantization format
Q2_K_S
Q2_K_S has a nominal rate of — bits per weight. We have no measured files carrying this label yet.
From the file· 86 files measured
Nominal bpw
—
from the block layout
Measured average
3.184
86 files
Range
2.701–14.149
varies by architecture
File sizes
0.03 GiB+
up to 74.13 GiB
Real files
one per model, most downloaded first
| Model | Size● | Effective bpw● | vs nominal● |
|---|---|---|---|
| Qwen-Image-Edit-2509 | 6.55 GiB | 2.752 | — |
| Codestral-22B-v0.1 | 7.12 GiB | 2.751 | — |
| DeepSeek-Coder-V2-Lite-Instruct | 6.01 GiB | 3.288 | — |
| Hermes-4-14B | 5.02 GiB | 2.920 | — |
| Qwen-Image-Edit | 6.55 GiB | 2.752 | — |
| Hermes-3-Llama-3.1-8B | 2.78 GiB | 2.978 | — |
| Qwen2-0.5B-Instruct | 0.31 GiB | 5.361 | — |
| Qwen2.5-Coder-32B | 10.70 GiB | 2.805 | — |
| Equinox-31B | 10.22 GiB | 2.808 | — |
| FLUX.1-Krea-dev | 3.83 GiB | 2.762 | — |
| gemma-2-27b-it | 9.06 GiB | 2.857 | — |
| Llama-3.1-70B | 22.79 GiB | 2.775 | — |
| glm-4-9b-chat | 3.69 GiB | 3.369 | — |
| Wan_2.2_ComfyUI_Repackaged | 8.86 GiB | — | — |
| Huihui-Qwen3-VL-30B-A3B-Instruct-abliterated | 9.80 GiB | 2.708 | — |
| gemma-4-12B | 4.38 GiB | 3.148 | — |
| Qwen2.5-Math-1.5B-Instruct | 0.60 GiB | 3.317 | — |
| Meta-Llama-3.1-8B-Instruct-abliterated | 2.78 GiB | 2.978 | — |
| Phi-3-mini-4k-instruct | 1.24 GiB | 2.779 | — |
| openchat-3.6-8b-20240522 | 2.78 GiB | 2.978 | — |
| DeepSeek-V2-Lite-Chat | 6.01 GiB | 3.288 | — |
| Hermes-3-Llama-3.1-70B | 22.79 GiB | 2.775 | — |
| Qwen3-Coder-REAP-25B-A3B | 8.01 GiB | 2.765 | — |
| Phi-3-medium-128k-instruct | 4.44 GiB | 2.734 | — |
| Ministral-3-8B-Instruct-2512-BF16 | 2.96 GiB | 2.853 | — |
●From the filewhat these mean