Is Rocinante-XL-16B-v1 or Qwen3.6-27B smaller?

Rocinante-XL-16B-v1 is smaller at Q4_K_M: 9.37 GiB against 15.41 GiB.

Which uses less memory at long context?

Qwen3.6-27B — its KV cache at 32K context is 3.4x smaller, because of how its attention layers cache.

Model comparison

Rocinante-XL-16B-v1 vs Qwen3.6-27B

At Q4_K_M, Rocinante-XL-16B-v1 is the smaller download — 10,063,921,216 bytes against 16,547,397,888. At long context the gap widens: Qwen3.6-27B's KV cache at 32K is 3.4× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Rocinante-XL-16B-v1	Qwen3.6-27B
Parameters	16.1B	27.8B
Architecture	llama	qwen35
Layers	54	64
Native context	131,072	262,144
Mixture of experts	no	no
Quantizations published	48	40
Smallest quantization	3.54 GiB	8.74 GiB
Q4_K_M	9.37 GiB	15.41 GiB
Licence	—	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Rocinante-XL-16B-v1	Qwen3.6-27B	Ratio
4,096	0.84 GiB	0.25 GiB	3.37×
8,192	1.69 GiB	0.50 GiB	3.37×
16,384	3.38 GiB	1.00 GiB	3.37×
32,768	6.75 GiB	2.00 GiB	3.37×
65,536	13.50 GiB	4.00 GiB	3.37×
131,072	27.00 GiB	8.00 GiB	3.37×