Is granite-3.2-2b-instruct or Qwen3-4B smaller?

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

Which uses less memory at long context?

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

Model comparison

granite-3.2-2b-instruct vs Qwen3-4B

At Q4_K_M, granite-3.2-2b-instruct is the smaller download — 1,545,296,256 bytes against 2,497,280,256. At long context the gap widens: granite-3.2-2b-instruct'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-3.2-2b-instruct	Qwen3-4B
Parameters	2.5B	4.0B
Architecture	granite	qwen3
Layers	40	36
Native context	131,072	40,960
Mixture of experts	no	no
Quantizations published	52	29
Smallest quantization	0.83 GiB	1.01 GiB
Q4_K_M	1.44 GiB	2.33 GiB
Licence	apache-2.0	—

KV cache by context

the term that decides long-context viability

Context	granite-3.2-2b-instruct	Qwen3-4B	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×