Is Qwen3-Coder-30B-A3B-Instruct or glm-4-9b-chat-abliterated smaller?

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

Which uses less memory at long context?

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

Model comparison

Qwen3-Coder-30B-A3B-Instruct vs glm-4-9b-chat-abliterated

At Q4_K_M, glm-4-9b-chat-abliterated is the smaller download — 6,250,927,200 bytes against 18,556,689,568. At long context the gap widens: Qwen3-Coder-30B-A3B-Instruct'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

	Qwen3-Coder-30B-A3B-Instruct	glm-4-9b-chat-abliterated
Parameters	30.5B	9.4B
Architecture	qwen3moe	chatglm
Layers	48	40
Native context	262,144	—
Mixture of experts	yes, 128 experts	no
Quantizations published	46	20
Smallest quantization	7.46 GiB	3.66 GiB
Q4_K_M	17.28 GiB	5.82 GiB
Licence	apache-2.0	other

KV cache by context

the term that decides long-context viability

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