Model comparison

ArrowMint-Gemma3-4B-YUKI-v0.1 vs Llama-3.1-8B-Instruct

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: ArrowMint-Gemma3-4B-YUKI-v0.1's KV cache at 32K is 5.0× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

ArrowMint-Gemma3-4B-YUKI-v0.1Llama-3.1-8B-Instruct
Parameters4.3B8.0B
Architecturegemma3llama
Layers3432
Native context131,072131,072
Mixture of expertsnono
Quantizations published2445
Smallest quantization1.06 GiB2.02 GiB
Q4_K_M4.58 GiB
Licencegemmallama3.1

KV cache by context

the term that decides long-context viability
ContextArrowMint-Gemma3-4B-YUKI-v0.1Llama-3.1-8B-InstructRatio
4,0960.25 GiB0.50 GiB2.02×
8,1920.33 GiB1.00 GiB3.07×
16,3840.48 GiB2.00 GiB4.15×
32,7680.79 GiB4.00 GiB5.03×
65,5361.42 GiB8.00 GiB5.63×
131,0722.67 GiB16.00 GiB5.99×