Is SuperHY3-abliterated-NVFP4 or GLM-5.2 smaller?

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

Which uses less memory at long context?

GLM-5.2 — its KV cache at 32K context is 3.6x smaller, because of how its attention layers cache.

Model comparison

SuperHY3-abliterated-NVFP4 vs GLM-5.2

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: GLM-5.2's KV cache at 32K is 3.6× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	SuperHY3-abliterated-NVFP4	GLM-5.2
Parameters	172B	753B
Architecture	hy-v3	glm-dsa
Layers	80	78
Native context	262,144	1,048,576
Mixture of experts	yes, 192 experts	yes, 256 experts
Quantizations published	1	20
Smallest quantization	95.10 GiB	169.33 GiB
Q4_K_M	—	—
Licence	apache-2.0	mit

KV cache by context

the term that decides long-context viability

Context	SuperHY3-abliterated-NVFP4	GLM-5.2	Ratio
4,096	1.25 GiB	0.34 GiB	3.65×
8,192	2.50 GiB	0.69 GiB	3.65×
16,384	5.00 GiB	1.37 GiB	3.65×
32,768	10.00 GiB	2.74 GiB	3.65×
65,536	20.00 GiB	5.48 GiB	3.65×
131,072	40.00 GiB	10.97 GiB	3.65×