Is Lilith-31B-v1.0 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 2.5x smaller, because of how its attention layers cache.

Model comparison

Lilith-31B-v1.0 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 2.5× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Lilith-31B-v1.0	gemma-4-12B-it-qat-q4_0-unquantized
Parameters	31.3B	12.0B
Architecture	gemma4	gemma4
Layers	60	48
Native context	262,144	262,144
Mixture of experts	no	no
Quantizations published	41	2
Smallest quantization	6.67 GiB	6.50 GiB
Q4_K_M	17.40 GiB	—
Licence	gemma	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Lilith-31B-v1.0	gemma-4-12B-it-qat-q4_0-unquantized	Ratio
4,096	1.80 GiB	0.72 GiB	2.50×
8,192	2.42 GiB	0.97 GiB	2.50×
16,384	3.67 GiB	1.47 GiB	2.50×
32,768	6.17 GiB	2.47 GiB	2.50×
65,536	11.17 GiB	4.47 GiB	2.50×
131,072	21.17 GiB	8.47 GiB	2.50×