Is Amaretto-3B or Qwen3-8B smaller?

Amaretto-3B is smaller at Q4_K_M: 2.00 GiB against 4.68 GiB.

Which uses less memory at long context?

Amaretto-3B — its KV cache at 32K context is 1.4x smaller, because of how its attention layers cache.

Model comparison

Amaretto-3B vs Qwen3-8B

At Q4_K_M, Amaretto-3B is the smaller download — 2,146,498,240 bytes against 5,027,783,488. At long context the gap widens: Amaretto-3B's KV cache at 32K is 1.4× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Amaretto-3B	Qwen3-8B
Parameters	4.3B	8.2B
Architecture	mistral3	qwen3
Layers	26	36
Native context	262,144	40,960
Mixture of experts	no	no
Quantizations published	36	51
Smallest quantization	0.87 GiB	2.12 GiB
Q4_K_M	2.00 GiB	4.68 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Amaretto-3B	Qwen3-8B	Ratio
4,096	0.41 GiB	0.56 GiB	1.38×
8,192	0.81 GiB	1.13 GiB	1.38×
16,384	1.63 GiB	2.25 GiB	1.38×
32,768	3.25 GiB	4.50 GiB	1.38×
65,536	6.50 GiB	9.00 GiB	1.38×
131,072	13.00 GiB	18.00 GiB	1.38×