Is gemma-4-12B-it-qat-q4_0-unquantized or Llama-3_3-Nemotron-Super-49B-v1 smaller?

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

Which uses less memory at long context?

gemma-4-12B-it-qat-q4_0-unquantized — its KV cache at 32K context is 32.4x smaller, because of how its attention layers cache.

Model comparison

gemma-4-12B-it-qat-q4_0-unquantized vs Llama-3_3-Nemotron-Super-49B-v1

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: gemma-4-12B-it-qat-q4_0-unquantized's KV cache at 32K is 32.4× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	gemma-4-12B-it-qat-q4_0-unquantized	Llama-3_3-Nemotron-Super-49B-v1
Parameters	12.0B	49.9B
Architecture	gemma4	deci
Layers	48	80
Native context	262,144	131,072
Mixture of experts	no	no
Quantizations published	2	66
Smallest quantization	6.50 GiB	10.27 GiB
Q4_K_M	—	28.14 GiB
Licence	apache-2.0	other

KV cache by context

the term that decides long-context viability

Context	gemma-4-12B-it-qat-q4_0-unquantized	Llama-3_3-Nemotron-Super-49B-v1	Ratio
4,096	0.72 GiB	10.00 GiB	13.91×
8,192	0.97 GiB	20.00 GiB	20.65×
16,384	1.47 GiB	40.00 GiB	27.23×
32,768	2.47 GiB	80.00 GiB	32.41×
65,536	4.47 GiB	160.00 GiB	35.80×
131,072	8.47 GiB	320.00 GiB	37.79×