Is Qwen3.6-27B or Huihui-Qwen3-Coder-Next-abliterated smaller?

Qwen3.6-27B is smaller at Q4_K_M: 15.41 GiB against 45.22 GiB.

Which uses less memory at long context?

Huihui-Qwen3-Coder-Next-abliterated — its KV cache at 32K context is 2.7x smaller, because of how its attention layers cache.

Model comparison

Qwen3.6-27B vs Huihui-Qwen3-Coder-Next-abliterated

At Q4_K_M, Qwen3.6-27B is the smaller download — 16,547,397,888 bytes against 48,556,632,000. At long context the gap widens: Huihui-Qwen3-Coder-Next-abliterated's KV cache at 32K is 2.7× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Qwen3.6-27B	Huihui-Qwen3-Coder-Next-abliterated
Parameters	27.8B	79.7B
Architecture	qwen35	qwen3next
Layers	64	48
Native context	262,144	262,144
Mixture of experts	no	yes, 512 experts
Quantizations published	40	73
Smallest quantization	8.74 GiB	15.07 GiB
Q4_K_M	15.41 GiB	45.22 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Qwen3.6-27B	Huihui-Qwen3-Coder-Next-abliterated	Ratio
4,096	0.25 GiB	0.09 GiB	2.67×
8,192	0.50 GiB	0.19 GiB	2.67×
16,384	1.00 GiB	0.38 GiB	2.67×
32,768	2.00 GiB	0.75 GiB	2.67×
65,536	4.00 GiB	1.50 GiB	2.67×
131,072	8.00 GiB	3.00 GiB	2.67×