Model comparison

Pixtral-12B-2409 vs gemma-4-12B-it

At Q4_K_M, gemma-4-12B-it is the smaller download — 7,121,861,440 bytes against 7,703,795,680.

From the file· summed bytes, KV per layer

Side by side

Pixtral-12B-2409gemma-4-12B-it
Parameters12.2B12.0B
Architecturellamagemma4
Layers48
Native context262,144
Mixture of expertsnono
Quantizations published1659
Smallest quantization4.77 GiB3.92 GiB
Q4_K_M7.17 GiB6.63 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextPixtral-12B-2409gemma-4-12B-itRatio
4,0960.72 GiB
8,1920.97 GiB
16,3841.47 GiB
32,7682.47 GiB
65,5364.47 GiB
131,0728.47 GiB