Is Legion-V2.1-LLaMa-70B 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?

gemma-4-12B-it-qat-q4_0-unquantized — its KV cache at 32K context is 4.1x smaller, because of how its attention layers cache.

Model comparison

Legion-V2.1-LLaMa-70B 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: gemma-4-12B-it-qat-q4_0-unquantized's KV cache at 32K is 4.1× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Legion-V2.1-LLaMa-70B	gemma-4-12B-it-qat-q4_0-unquantized
Parameters	70.6B	12.0B
Architecture	llama	gemma4
Layers	80	48
Native context	131,072	262,144
Mixture of experts	no	no
Quantizations published	45	2
Smallest quantization	14.29 GiB	6.50 GiB
Q4_K_M	39.60 GiB	—
Licence	llama3.3	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Legion-V2.1-LLaMa-70B	gemma-4-12B-it-qat-q4_0-unquantized	Ratio
4,096	1.25 GiB	0.72 GiB	1.74×
8,192	2.50 GiB	0.97 GiB	2.58×
16,384	5.00 GiB	1.47 GiB	3.40×
32,768	10.00 GiB	2.47 GiB	4.05×
65,536	20.00 GiB	4.47 GiB	4.48×
131,072	40.00 GiB	8.47 GiB	4.72×