Is GLM-4.7-Flash or gemma-4-12B-it-qat-q4_0-unquantized smaller?

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

Which uses less memory at long context?

GLM-4.7-Flash — its KV cache at 32K context is 1.5x smaller, because of how its attention layers cache.

Model comparison

GLM-4.7-Flash vs gemma-4-12B-it-qat-q4_0-unquantized

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: GLM-4.7-Flash'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

	GLM-4.7-Flash	gemma-4-12B-it-qat-q4_0-unquantized
Parameters	31.2B	12.0B
Architecture	deepseek2	gemma4
Layers	47	48
Native context	202,752	262,144
Mixture of experts	yes, 64 experts	no
Quantizations published	26	2
Smallest quantization	7.76 GiB	6.50 GiB
Q4_K_M	16.89 GiB	—
Licence	—	apache-2.0

KV cache by context

the term that decides long-context viability

Context	GLM-4.7-Flash	gemma-4-12B-it-qat-q4_0-unquantized	Ratio
4,096	0.21 GiB	0.72 GiB	3.48×
8,192	0.41 GiB	0.97 GiB	2.35×
16,384	0.83 GiB	1.47 GiB	1.78×
32,768	1.65 GiB	2.47 GiB	1.49×
65,536	3.30 GiB	4.47 GiB	1.35×
131,072	6.61 GiB	8.47 GiB	1.28×