Is Gemma-3-4B-VL-it-Gemini-Pro-Heretic-Uncensored-Thinking or gemma-4-E2B-it smaller?

Gemma-3-4B-VL-it-Gemini-Pro-Heretic-Uncensored-Thinking is smaller at Q4_K_M: 2.32 GiB against 2.89 GiB.

Which uses less memory at long context?

gemma-4-E2B-it — its KV cache at 32K context is 3.2x smaller, because of how its attention layers cache.

Model comparison

Gemma-3-4B-VL-it-Gemini-Pro-Heretic-Uncensored-Thinking vs gemma-4-E2B-it

At Q4_K_M, Gemma-3-4B-VL-it-Gemini-Pro-Heretic-Uncensored-Thinking is the smaller download — 2,489,895,200 bytes against 3,106,738,272. At long context the gap widens: gemma-4-E2B-it's KV cache at 32K is 3.2× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Gemma-3-4B-VL-it-Gemini-Pro-Heretic-Uncensored-Thinking	gemma-4-E2B-it
Parameters	4.3B	5.1B
Architecture	gemma3	gemma4
Layers	34	35
Native context	131,072	131,072
Mixture of experts	no	no
Quantizations published	12	54
Smallest quantization	1.61 GiB	2.13 GiB
Q4_K_M	2.32 GiB	2.89 GiB
Licence	apache-2.0	apache-2.0

KV cache by context

the term that decides long-context viability

Context	Gemma-3-4B-VL-it-Gemini-Pro-Heretic-Uncensored-Thinking	gemma-4-E2B-it	Ratio
4,096	0.25 GiB	0.05 GiB	4.54×
8,192	0.33 GiB	0.08 GiB	3.98×
16,384	0.48 GiB	0.14 GiB	3.53×
32,768	0.79 GiB	0.25 GiB	3.23×
65,536	1.42 GiB	0.46 GiB	3.05×
131,072	2.67 GiB	0.90 GiB	2.96×