Is llm-jp-4-32b-a3b-thinking or Qwen3-Coder-30B-A3B-Instruct smaller?

Qwen3-Coder-30B-A3B-Instruct is smaller at Q4_K_M: 17.28 GiB against 19.93 GiB.

Which uses less memory at long context?

llm-jp-4-32b-a3b-thinking — its KV cache at 32K context is 1.5x smaller, because of how its attention layers cache.

Model comparison

llm-jp-4-32b-a3b-thinking 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 21,400,590,016. At long context the gap widens: llm-jp-4-32b-a3b-thinking's KV cache at 32K is 1.5× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	llm-jp-4-32b-a3b-thinking	Qwen3-Coder-30B-A3B-Instruct
Parameters	32.1B	30.5B
Architecture	qwen3moe	qwen3moe
Layers	32	48
Native context	65,536	262,144
Mixture of experts	yes, 128 experts	yes, 128 experts
Quantizations published	11	46
Smallest quantization	15.60 GiB	7.46 GiB
Q4_K_M	19.93 GiB	17.28 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	llm-jp-4-32b-a3b-thinking	Qwen3-Coder-30B-A3B-Instruct	Ratio
4,096	0.25 GiB	0.38 GiB	1.50×
8,192	0.50 GiB	0.75 GiB	1.50×
16,384	1.00 GiB	1.50 GiB	1.50×
32,768	2.00 GiB	3.00 GiB	1.50×
65,536	4.00 GiB	6.00 GiB	1.50×
131,072	8.00 GiB	12.00 GiB	1.50×