Quantization format

I1_IQ2_XS

I1_IQ2_XS has a nominal rate of — bits per weight. We have no measured files carrying this label yet.

From the file· 957 files measured

Nominal bpw

—

from the block layout

Measured average

2.647

957 files

Range

1.546–5.613

varies by architecture

File sizes

0.06 GiB+

up to 107.87 GiB

Real files

one per model, most downloaded first

Model	Size●	Effective bpw●	vs nominal●
Qwopus3.6-35B-A3B-v1	9.79 GiB	2.338	—
Qwen3-VL-8B-Instruct-abliterated-v1	2.51 GiB	2.460	—
gemma-4-26B-A4B-it-ultra-uncensored-heretic	9.14 GiB	3.043	—
Sugoi-14B-Ultra-HF	4.38 GiB	2.548	—
gemma-4-12b-heretic-abliterated	3.62 GiB	2.601	—
FastContext-1.0-4B-SFT	1.26 GiB	2.693	—
gemma-4-31B-it-The-DECKARD-HERETIC-UNCENSORED-Thinking	8.88 GiB	2.438	—
Qwen3.6-35B-A3B-Claude-4.6-Opus-Reasoning-Distilled	9.79 GiB	2.338	—
Llama3.3-8B-Instruct-Thinking-Heretic-Uncensored-Claude-4.5-Opus-High-Reasoning	2.43 GiB	2.596	—
Mistral-7B-Instruct-v0.2	2.05 GiB	2.428	—
Huihui-Qwen3-Coder-Next-abliterated	21.82 GiB	2.353	—
DeepSeek-R1-Distill-Llama-70B-abliterated	19.69 GiB	2.397	—
gemma-4-26B-A4B-it-heretic	9.14 GiB	3.043	—
gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-uncensored-heretic	3.62 GiB	2.601	—
Qwen3.5-122B-A10B-heretic-v2	33.43 GiB	2.351	—
Llama-3.3-70B-Instruct-abliterated	19.69 GiB	2.397	—
Qwable-9B-Claude-Fable-5	3.06 GiB	2.793	—
EVA-Qwen2.5-14B-v0.2	4.38 GiB	2.548	—
Prosopon-31B	9.31 GiB	2.446	—
Dirty-Muse-Writer-v01-Uncensored-Erotica-NSFW	2.86 GiB	2.655	—
Gemma-4-26B-A4B-StyleTune-V2	9.37 GiB	3.031	—
Qwen2.5-VL-7B-Instruct-abliterated	2.30 GiB	2.382	—
Qwen3.5-9B-Claude-4.6-Opus-Deckard-V4.2-Uncensored-Heretic-Thinking	3.06 GiB	2.793	—
MythoMax-L2-13b	3.82 GiB	2.524	—
Mistral-7B-Instruct-v0.1	2.05 GiB	2.428	—

●From the filewhat these mean