Is embeddinggemma-300m or Llama-3.2-1B-Instruct smaller?

Both publish quantizations; compare the table for the exact sizes.

Which uses less memory at long context?

embeddinggemma-300m — its KV cache at 32K context is 6.9x smaller, because of how its attention layers cache.

Model comparison

embeddinggemma-300m vs Llama-3.2-1B-Instruct

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: embeddinggemma-300m's KV cache at 32K is 6.9× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	embeddinggemma-300m	Llama-3.2-1B-Instruct
Parameters	303M	1.2B
Architecture	gemma-embedding	llama
Layers	24	16
Native context	2,048	131,072
Mixture of experts	no	no
Quantizations published	10	39
Smallest quantization	0.26 GiB	0.39 GiB
Q4_K_M	—	0.75 GiB
Licence	—	—

KV cache by context

the term that decides long-context viability

Context	embeddinggemma-300m	Llama-3.2-1B-Instruct	Ratio
4,096	0.04 GiB	0.13 GiB	3.56×
8,192	0.05 GiB	0.25 GiB	4.92×
16,384	0.08 GiB	0.50 GiB	6.10×
32,768	0.14 GiB	1.00 GiB	6.92×
65,536	0.27 GiB	2.00 GiB	7.42×
131,072	0.52 GiB	4.00 GiB	7.70×