Is HarmonicHarlequin_v5-20B or gemma-4-E4B-it smaller?

gemma-4-E4B-it is smaller at Q4_K_M: 4.64 GiB against 18.71 GiB.

Which uses less memory at long context?

gemma-4-E4B-it — its KV cache at 32K context is 128.5x smaller, because of how its attention layers cache.

Model comparison

HarmonicHarlequin_v5-20B vs gemma-4-E4B-it

At Q4_K_M, gemma-4-E4B-it is the smaller download — 4,977,171,584 bytes against 20,087,800,544. At long context the gap widens: gemma-4-E4B-it's KV cache at 32K is 128.5× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	HarmonicHarlequin_v5-20B	gemma-4-E4B-it
Parameters	33.3B	8.0B
Architecture	llama	gemma4
Layers	104	42
Native context	4,096	131,072
Mixture of experts	no	no
Quantizations published	35	36
Smallest quantization	6.77 GiB	3.30 GiB
Q4_K_M	18.71 GiB	4.64 GiB
Licence	—	apache-2.0

KV cache by context

the term that decides long-context viability

Context	HarmonicHarlequin_v5-20B	gemma-4-E4B-it	Ratio
4,096	8.13 GiB	0.12 GiB	66.03×
8,192	16.25 GiB	0.18 GiB	91.43×
16,384	32.50 GiB	0.29 GiB	113.20×
32,768	65.00 GiB	0.51 GiB	128.49×
65,536	130.00 GiB	0.94 GiB	137.81×
131,072	260.00 GiB	1.82 GiB	142.99×