Is Mellum2-12B-A2.5B-Thinking or Qwen3.6-27B smaller?

Mellum2-12B-A2.5B-Thinking is smaller at Q4_K_M: 7.52 GiB against 15.41 GiB.

Which uses less memory at long context?

Mellum2-12B-A2.5B-Thinking — its KV cache at 32K context is 4.0x smaller, because of how its attention layers cache.

Model comparison

Mellum2-12B-A2.5B-Thinking vs Qwen3.6-27B

At Q4_K_M, Mellum2-12B-A2.5B-Thinking is the smaller download — 8,071,295,040 bytes against 16,547,397,888. At long context the gap widens: Mellum2-12B-A2.5B-Thinking's KV cache at 32K is 4.0× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Mellum2-12B-A2.5B-Thinking	Qwen3.6-27B
Parameters	12.1B	27.8B
Architecture	mellum	qwen35
Layers	28	64
Native context	131,072	262,144
Mixture of experts	yes, 64 experts	no
Quantizations published	3	40
Smallest quantization	7.52 GiB	8.74 GiB
Q4_K_M	7.52 GiB	15.41 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Mellum2-12B-A2.5B-Thinking	Qwen3.6-27B	Ratio
4,096	0.12 GiB	0.25 GiB	2.15×
8,192	0.17 GiB	0.50 GiB	2.93×
16,384	0.28 GiB	1.00 GiB	3.57×
32,768	0.50 GiB	2.00 GiB	4.01×
65,536	0.94 GiB	4.00 GiB	4.27×
131,072	1.81 GiB	8.00 GiB	4.42×