Quantization format
IQ3_S
IQ3_S has a nominal rate of — bits per weight. We have no measured files carrying this label yet.
From the file· 235 files measured
Nominal bpw
—
from the block layout
Measured average
3.867
235 files
Range
2.884–20.569
varies by architecture
File sizes
0.03 GiB+
up to 377.51 GiB
Real files
one per model, most downloaded first
| Model | Size● | Effective bpw● | vs nominal● |
|---|---|---|---|
| Qwen-Image-Edit-2509 | 22.43 GiB | 9.430 | — |
| Qwen3-0.6B-Base | 0.36 GiB | 5.234 | — |
| Mathstral-7B-v0.1 | 2.97 GiB | 3.517 | — |
| Krea-2-Raw | 5.14 GiB | 3.441 | — |
| Codestral-22B-v0.1 | 9.02 GiB | 3.484 | — |
| Mistral-7B-v0.1 | 2.96 GiB | 3.516 | — |
| Nanbeige4.2-3B | 1.87 GiB | 3.846 | — |
| MN-12B-Mag-Mell-R1 | 5.18 GiB | 3.633 | — |
| DeepSeek-Coder-V2-Lite-Instruct | 6.97 GiB | 3.814 | — |
| Qwen3.6-12B-IQ-Ultra-Heretic-Uncensored-Thinking-V2-Hightop | 5.27 GiB | 3.729 | — |
| Llama-3.2-3B-Instruct-uncensored | 1.59 GiB | 3.798 | — |
| Gemma-4-31B-StyleTune | 12.22 GiB | 3.213 | — |
| gemma-2-2b-it-abliterated | 1.27 GiB | 4.164 | — |
| dolphin-2.9-llama3-8b | 3.43 GiB | 3.668 | — |
| Hermes-4-14B | 6.23 GiB | 3.621 | — |
| Qwen-Image-Edit | 8.36 GiB | 3.516 | — |
| NEXUS-Medical | 0.71 GiB | 3.951 | — |
| L3-8B-Stheno-v3.2 | 3.43 GiB | 3.668 | — |
| Meta-Llama-3-8B | 3.43 GiB | 3.668 | — |
| MN-Violet-Lotus-12B | 5.18 GiB | 3.633 | — |
| MythoMax-L2-Kimiko-v2-13b | 5.45 GiB | 3.597 | — |
| Hermes-3-Llama-3.1-8B | 3.43 GiB | 3.668 | — |
| Qwen2-0.5B-Instruct | 0.32 GiB | 5.483 | — |
| Mixtral-8x7B-Instruct-v0.1 | 19.03 GiB | 3.500 | — |
| Qwen2.5-Coder-32B | 13.45 GiB | 3.525 | — |
●From the filewhat these mean