Is Llama-3.2-11B-Vision-Instruct or Qwen3.6-27B smaller?

Llama-3.2-11B-Vision-Instruct is smaller at Q4_K_M: 5.55 GiB against 15.41 GiB.

Which uses less memory at long context?

Qwen3.6-27B — its KV cache at 32K context is 2.5x smaller, because of how its attention layers cache.

Model comparison

Llama-3.2-11B-Vision-Instruct vs Qwen3.6-27B

At Q4_K_M, Llama-3.2-11B-Vision-Instruct is the smaller download — 5,963,057,216 bytes against 16,547,397,888. At long context the gap widens: Qwen3.6-27B's KV cache at 32K is 2.5× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Llama-3.2-11B-Vision-Instruct	Qwen3.6-27B
Parameters	10.7B	27.8B
Architecture	mllama	qwen35
Layers	40	64
Native context	131,072	262,144
Mixture of experts	no	no
Quantizations published	9	40
Smallest quantization	5.55 GiB	8.74 GiB
Q4_K_M	5.55 GiB	15.41 GiB
Licence	llama3.2	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Llama-3.2-11B-Vision-Instruct	Qwen3.6-27B	Ratio
4,096	0.63 GiB	0.25 GiB	2.50×
8,192	1.25 GiB	0.50 GiB	2.50×
16,384	2.50 GiB	1.00 GiB	2.50×
32,768	5.00 GiB	2.00 GiB	2.50×
65,536	10.00 GiB	4.00 GiB	2.50×
131,072	20.00 GiB	8.00 GiB	2.50×