Is granite-4.0-h-small or Qwen3-Coder-30B-A3B-Instruct smaller?

Qwen3-Coder-30B-A3B-Instruct is smaller at Q4_K_M: 17.28 GiB against 18.14 GiB.

Which uses less memory at long context?

granite-4.0-h-small — its KV cache at 32K context is 6.0x smaller, because of how its attention layers cache.

Model comparison

granite-4.0-h-small vs Qwen3-Coder-30B-A3B-Instruct

At Q4_K_M, Qwen3-Coder-30B-A3B-Instruct is the smaller download — 18,556,689,568 bytes against 19,476,621,984. At long context the gap widens: granite-4.0-h-small's KV cache at 32K is 6.0× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	granite-4.0-h-small	Qwen3-Coder-30B-A3B-Instruct
Parameters	32.2B	30.5B
Architecture	granitehybrid	qwen3moe
Layers	40	48
Native context	131,072	262,144
Mixture of experts	yes, 72 experts	yes, 128 experts
Quantizations published	41	46
Smallest quantization	7.59 GiB	7.46 GiB
Q4_K_M	18.14 GiB	17.28 GiB
Licence	—	apache-2.0

KV cache by context

the term that decides long-context viability

Context	granite-4.0-h-small	Qwen3-Coder-30B-A3B-Instruct	Ratio
4,096	0.06 GiB	0.38 GiB	5.99×
8,192	0.13 GiB	0.75 GiB	5.99×
16,384	0.25 GiB	1.50 GiB	6.00×
32,768	0.50 GiB	3.00 GiB	6.00×
65,536	1.00 GiB	6.00 GiB	6.00×
131,072	2.00 GiB	12.00 GiB	6.00×