Is Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-MTP or grug-35b smaller?

grug-35b is smaller at Q4_K_M: 19.71 GiB against 34.04 GiB.

Which uses less memory at long context?

grug-35b — its KV cache at 32K context is 3.2x smaller, because of how its attention layers cache.

Model comparison

Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-MTP vs grug-35b

At Q4_K_M, grug-35b is the smaller download — 21,166,758,528 bytes against 36,545,830,912. At long context the gap widens: grug-35b's KV cache at 32K is 3.2× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-MTP	grug-35b
Parameters	27.8B	35.1B
Architecture	qwen35	qwen35moe
Layers	64	40
Native context	262,144	262,144
Mixture of experts	no	yes, 256 experts
Quantizations published	34	30
Smallest quantization	10.12 GiB	9.11 GiB
Q4_K_M	34.04 GiB	19.71 GiB
Licence	apache-2.0	mit

KV cache by context

the term that decides long-context viability

Context	Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-MTP	grug-35b	Ratio
4,096	0.25 GiB	0.08 GiB	3.20×
8,192	0.50 GiB	0.16 GiB	3.20×
16,384	1.00 GiB	0.31 GiB	3.20×
32,768	2.00 GiB	0.63 GiB	3.20×
65,536	4.00 GiB	1.25 GiB	3.20×
131,072	8.00 GiB	2.50 GiB	3.20×