Is gemma-4-31B-it-NVFP4 or gemma-4-26B-A4B-it smaller?

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

Which uses less memory at long context?

gemma-4-26B-A4B-it — its KV cache at 32K context is 4.0x smaller, because of how its attention layers cache.

Model comparison

gemma-4-31B-it-NVFP4 vs gemma-4-26B-A4B-it

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: gemma-4-26B-A4B-it's KV cache at 32K is 4.0× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	gemma-4-31B-it-NVFP4	gemma-4-26B-A4B-it
Parameters	19.9B	26.5B
Architecture	gemma4	gemma4
Layers	60	30
Native context	262,144	262,144
Mixture of experts	no	yes, 128 experts
Quantizations published	1	46
Smallest quantization	20.61 GiB	8.99 GiB
Q4_K_M	—	15.64 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	gemma-4-31B-it-NVFP4	gemma-4-26B-A4B-it	Ratio
4,096	1.80 GiB	0.45 GiB	4.00×
8,192	2.42 GiB	0.61 GiB	4.00×
16,384	3.67 GiB	0.92 GiB	4.00×
32,768	6.17 GiB	1.54 GiB	4.00×
65,536	11.17 GiB	2.79 GiB	4.00×
131,072	21.17 GiB	5.29 GiB	4.00×