Is Qwen3-VL-32B-Instruct or gemma-4-26B-A4B-it smaller?

gemma-4-26B-A4B-it is smaller at Q4_K_M: 15.64 GiB against 18.40 GiB.

Which uses less memory at long context?

gemma-4-26B-A4B-it — its KV cache at 32K context is 5.2x smaller, because of how its attention layers cache.

Model comparison

Qwen3-VL-32B-Instruct vs gemma-4-26B-A4B-it

At Q4_K_M, gemma-4-26B-A4B-it is the smaller download — 16,796,017,248 bytes against 19,762,150,432. At long context the gap widens: gemma-4-26B-A4B-it's KV cache at 32K is 5.2× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Qwen3-VL-32B-Instruct	gemma-4-26B-A4B-it
Parameters	33.4B	26.5B
Architecture	qwen3vl	gemma4
Layers	64	30
Native context	262,144	262,144
Mixture of experts	no	yes, 128 experts
Quantizations published	48	46
Smallest quantization	7.21 GiB	8.99 GiB
Q4_K_M	18.40 GiB	15.64 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Qwen3-VL-32B-Instruct	gemma-4-26B-A4B-it	Ratio
4,096	1.00 GiB	0.45 GiB	2.23×
8,192	2.00 GiB	0.61 GiB	3.30×
16,384	4.00 GiB	0.92 GiB	4.36×
32,768	8.00 GiB	1.54 GiB	5.18×
65,536	16.00 GiB	2.79 GiB	5.73×
131,072	32.00 GiB	5.29 GiB	6.05×