Which uses less memory at long context?

Qwen3.6-27B — its KV cache at 32K context is 2.1x smaller, because of how its attention layers cache.

Model comparison

Qwen3-42B-A3B-2507-Thinking-Abliterated-uncensored-TOTAL-RECALL-v2-Medium-MASTER-CODER vs Qwen3.6-27B

Q: Is Qwen3-42B-A3B-2507-Thinking-Abliterated-uncensored-TOTAL-RECALL-v2-Medium-MASTER-CODER or Qwen3.6-27B smaller?

Qwen3.6-27B is smaller at Q4_K_M: 15.41 GiB against 23.94 GiB.

At Q4_K_M, Qwen3.6-27B is the smaller download — 16,547,397,888 bytes against 25,703,283,040. At long context the gap widens: Qwen3.6-27B's KV cache at 32K is 2.1× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Qwen3-42B-A3B-2507-Thinking-Abliterated-uncensored-TOTAL-RECALL-v2-Medium-MASTER-CODER	Qwen3.6-27B
Parameters	42.4B	27.8B
Architecture	qwen3moe	qwen35
Layers	67	64
Native context	262,144	262,144
Mixture of experts	yes, 128 experts	no
Quantizations published	34	40
Smallest quantization	8.21 GiB	8.74 GiB
Q4_K_M	23.94 GiB	15.41 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Qwen3-42B-A3B-2507-Thinking-Abliterated-uncensored-TOTAL-RECALL-v2-Medium-MASTER-CODER	Qwen3.6-27B	Ratio
4,096	0.52 GiB	0.25 GiB	2.09×
8,192	1.05 GiB	0.50 GiB	2.09×
16,384	2.09 GiB	1.00 GiB	2.09×
32,768	4.19 GiB	2.00 GiB	2.09×
65,536	8.38 GiB	4.00 GiB	2.09×
131,072	16.75 GiB	8.00 GiB	2.09×