Model comparison

Gemma-3-4B-VL-it-Gemini-Pro-Heretic-Uncensored-Thinking vs gemma-4-12B-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 7,121,861,440. At long context the gap widens: Gemma-3-4B-VL-it-Gemini-Pro-Heretic-Uncensored-Thinking's KV cache at 32K is 3.1× 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-Thinkinggemma-4-12B-it
Parameters4.3B12.0B
Architecturegemma3gemma4
Layers3448
Native context131,072262,144
Mixture of expertsnono
Quantizations published1259
Smallest quantization1.61 GiB3.92 GiB
Q4_K_M2.32 GiB6.63 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextGemma-3-4B-VL-it-Gemini-Pro-Heretic-Uncensored-Thinkinggemma-4-12B-itRatio
4,0960.25 GiB0.72 GiB2.90×
8,1920.33 GiB0.97 GiB2.97×
16,3840.48 GiB1.47 GiB3.04×
32,7680.79 GiB2.47 GiB3.11×
65,5361.42 GiB4.47 GiB3.15×
131,0722.67 GiB8.47 GiB3.17×
Gemma-3-4B-VL-it-Gemini-Pro-Heretic-Uncensored-Thinking vs gemma-4-12B-it — size, memory and hardware fit — ossmodeldb