Is gemma-2-2b-it or Llama-3.2-1B-Instruct smaller?

Llama-3.2-1B-Instruct is smaller at Q4_K_M: 0.75 GiB against 1.59 GiB.

Which uses less memory at long context?

Llama-3.2-1B-Instruct — its KV cache at 32K context is 1.9x smaller, because of how its attention layers cache.

Model comparison

gemma-2-2b-it vs Llama-3.2-1B-Instruct

At Q4_K_M, Llama-3.2-1B-Instruct is the smaller download — 807,690,688 bytes against 1,708,582,752. At long context the gap widens: Llama-3.2-1B-Instruct's KV cache at 32K is 1.9× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	gemma-2-2b-it	Llama-3.2-1B-Instruct
Parameters	2.6B	1.2B
Architecture	gemma2	llama
Layers	26	16
Native context	8,192	131,072
Mixture of experts	no	no
Quantizations published	34	39
Smallest quantization	0.78 GiB	0.39 GiB
Q4_K_M	1.59 GiB	0.75 GiB
Licence	gemma	—

KV cache by context

the term that decides long-context viability

Context	gemma-2-2b-it	Llama-3.2-1B-Instruct	Ratio
4,096	0.41 GiB	0.13 GiB	3.25×
8,192	0.63 GiB	0.25 GiB	2.54×
16,384	1.04 GiB	0.50 GiB	2.08×
32,768	1.85 GiB	1.00 GiB	1.85×
65,536	3.48 GiB	2.00 GiB	1.74×
131,072	6.73 GiB	4.00 GiB	1.68×