Model comparison

granite-4.0-h-small vs gemma-4-E4B-it

At Q4_K_M, gemma-4-E4B-it is the smaller download — 4,977,171,584 bytes against 19,476,621,984.

From the file· summed bytes, KV per layer

Side by side

granite-4.0-h-smallgemma-4-E4B-it
Parameters32.2B8.0B
Architecturegranitehybridgemma4
Layers4042
Native context131,072131,072
Mixture of expertsyes, 72 expertsno
Quantizations published4136
Smallest quantization7.59 GiB3.30 GiB
Q4_K_M18.14 GiB4.64 GiB
Licenceapache-2.0

KV cache by context

the term that decides long-context viability
Contextgranite-4.0-h-smallgemma-4-E4B-itRatio
4,0960.06 GiB0.12 GiB1.96×
8,1920.13 GiB0.18 GiB1.42×
16,3840.25 GiB0.29 GiB1.15×
32,7680.50 GiB0.51 GiB1.01×
65,5361.00 GiB0.94 GiB1.06×
131,0722.00 GiB1.82 GiB1.10×