Is glm-4-9b-chat or Qwen3-30B-A3B-Thinking-2507 smaller?

glm-4-9b-chat is smaller at Q4_K_M: 5.82 GiB against 17.28 GiB.

Which uses less memory at long context?

Qwen3-30B-A3B-Thinking-2507 — its KV cache at 32K context is 6.7x smaller, because of how its attention layers cache.

Model comparison

glm-4-9b-chat vs Qwen3-30B-A3B-Thinking-2507

At Q4_K_M, glm-4-9b-chat is the smaller download — 6,250,926,848 bytes against 18,556,685,824. At long context the gap widens: Qwen3-30B-A3B-Thinking-2507's KV cache at 32K is 6.7× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	glm-4-9b-chat	Qwen3-30B-A3B-Thinking-2507
Parameters	9.4B	30.5B
Architecture	chatglm	qwen3moe
Layers	40	48
Native context	—	262,144
Mixture of experts	no	yes, 128 experts
Quantizations published	58	51
Smallest quantization	2.89 GiB	7.05 GiB
Q4_K_M	5.82 GiB	17.28 GiB
Licence	other	apache-2.0

KV cache by context

the term that decides long-context viability

Context	glm-4-9b-chat	Qwen3-30B-A3B-Thinking-2507	Ratio
4,096	2.50 GiB	0.38 GiB	6.67×
8,192	5.00 GiB	0.75 GiB	6.67×
16,384	10.00 GiB	1.50 GiB	6.67×
32,768	20.00 GiB	3.00 GiB	6.67×
65,536	40.00 GiB	6.00 GiB	6.67×
131,072	80.00 GiB	12.00 GiB	6.67×