Is granite-vision-3.2-2b or Qwen3-VL-4B-Instruct smaller?

granite-vision-3.2-2b is smaller at Q4_K_M: 1.44 GiB against 2.33 GiB.

Which uses less memory at long context?

granite-vision-3.2-2b — its KV cache at 32K context is 1.8x smaller, because of how its attention layers cache.

Model comparison

granite-vision-3.2-2b vs Qwen3-VL-4B-Instruct

At Q4_K_M, granite-vision-3.2-2b is the smaller download — 1,545,294,400 bytes against 2,497,281,568. At long context the gap widens: granite-vision-3.2-2b's KV cache at 32K is 1.8× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	granite-vision-3.2-2b	Qwen3-VL-4B-Instruct
Parameters	3.0B	4.4B
Architecture	granite	qwen3vl
Layers	40	36
Native context	131,072	262,144
Mixture of experts	no	no
Quantizations published	30	26
Smallest quantization	0.83 GiB	1.01 GiB
Q4_K_M	1.44 GiB	2.33 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	granite-vision-3.2-2b	Qwen3-VL-4B-Instruct	Ratio
4,096	0.31 GiB	0.56 GiB	1.80×
8,192	0.63 GiB	1.13 GiB	1.80×
16,384	1.25 GiB	2.25 GiB	1.80×
32,768	2.50 GiB	4.50 GiB	1.80×
65,536	5.00 GiB	9.00 GiB	1.80×
131,072	10.00 GiB	18.00 GiB	1.80×