Is Kimi-K2-Instruct or GLM-5.2 smaller?

Both publish quantizations; compare the table for the exact sizes.

Which uses less memory at long context?

Kimi-K2-Instruct — its KV cache at 32K context is 1.3x smaller, because of how its attention layers cache.

Model comparison

Kimi-K2-Instruct vs GLM-5.2

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: Kimi-K2-Instruct's KV cache at 32K is 1.3× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Kimi-K2-Instruct	GLM-5.2
Parameters	1026B	753B
Architecture	deepseek2	glm-dsa
Layers	61	78
Native context	131,072	1,048,576
Mixture of experts	yes, 384 experts	yes, 256 experts
Quantizations published	42	20
Smallest quantization	226.87 GiB	169.33 GiB
Q4_K_M	578.15 GiB	—
Licence	other	mit

KV cache by context

the term that decides long-context viability

Context	Kimi-K2-Instruct	GLM-5.2	Ratio
4,096	0.27 GiB	0.34 GiB	1.28×
8,192	0.54 GiB	0.69 GiB	1.28×
16,384	1.07 GiB	1.37 GiB	1.28×
32,768	2.14 GiB	2.74 GiB	1.28×
65,536	4.29 GiB	5.48 GiB	1.28×
131,072	8.58 GiB	10.97 GiB	1.28×