Is archangel_sft-kto_llama30b or gemma-4-E4B-it smaller?

gemma-4-E4B-it is smaller at Q4_K_M: 4.64 GiB against 18.27 GiB.

Which uses less memory at long context?

gemma-4-E4B-it — its KV cache at 32K context is 96.4x smaller, because of how its attention layers cache.

Model comparison

archangel_sft-kto_llama30b vs gemma-4-E4B-it

At Q4_K_M, gemma-4-E4B-it is the smaller download — 4,977,171,584 bytes against 19,621,140,512. At long context the gap widens: gemma-4-E4B-it's KV cache at 32K is 96.4× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	archangel_sft-kto_llama30b	gemma-4-E4B-it
Parameters	32.5B	8.0B
Architecture	llama	gemma4
Layers	60	42
Native context	2,048	131,072
Mixture of experts	no	no
Quantizations published	35	36
Smallest quantization	6.63 GiB	3.30 GiB
Q4_K_M	18.27 GiB	4.64 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	archangel_sft-kto_llama30b	gemma-4-E4B-it	Ratio
4,096	6.09 GiB	0.12 GiB	49.52×
8,192	12.19 GiB	0.18 GiB	68.57×
16,384	24.38 GiB	0.29 GiB	84.90×
32,768	48.75 GiB	0.51 GiB	96.37×
65,536	97.50 GiB	0.94 GiB	103.35×
131,072	195.00 GiB	1.82 GiB	107.24×