Model comparison

gemma-4-E4B-it-assistant vs SmolLM2-135M-Instruct

At Q4_K_M, gemma-4-E4B-it-assistant is the smaller download — 77,231,200 bytes against 105,454,144. At long context the gap widens: gemma-4-E4B-it-assistant's KV cache at 32K is 10.3× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

gemma-4-E4B-it-assistantSmolLM2-135M-Instruct
Parameters79M135M
Architecturegemma4-assistantllama
Layers430
Native context131,0728,192
Mixture of expertsnono
Quantizations published941
Smallest quantization0.07 GiB0.08 GiB
Q4_K_M0.07 GiB0.10 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
Contextgemma-4-E4B-it-assistantSmolLM2-135M-InstructRatio
4,0960.01 GiB0.09 GiB6.43×
8,1920.02 GiB0.18 GiB8.18×
16,3840.04 GiB0.35 GiB9.47×
32,7680.07 GiB0.70 GiB10.29×
65,5360.13 GiB1.41 GiB10.75×
131,0720.26 GiB2.81 GiB10.99×