Is embeddinggemma-300m or Qwen2.5-1.5B-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 6.1x smaller, because of how its attention layers cache.

Model comparison

embeddinggemma-300m vs Qwen2.5-1.5B-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 6.1× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	embeddinggemma-300m	Qwen2.5-1.5B-Instruct
Parameters	303M	1.5B
Architecture	gemma-embedding	qwen2
Layers	24	28
Native context	2,048	32,768
Mixture of experts	no	no
Quantizations published	10	32
Smallest quantization	0.26 GiB	0.56 GiB
Q4_K_M	—	0.92 GiB
Licence	—	apache-2.0

KV cache by context

the term that decides long-context viability

Context	embeddinggemma-300m	Qwen2.5-1.5B-Instruct	Ratio
4,096	0.04 GiB	0.11 GiB	3.11×
8,192	0.05 GiB	0.22 GiB	4.31×
16,384	0.08 GiB	0.44 GiB	5.33×
32,768	0.14 GiB	0.88 GiB	6.05×
65,536	0.27 GiB	1.75 GiB	6.49×
131,072	0.52 GiB	3.50 GiB	6.74×