Model comparison

gemma-4-26B-A4B-it vs Ministral-3-8B-Reasoning-2512-heretic

At Q4_K_M, Ministral-3-8B-Reasoning-2512-heretic is the smaller download — 5,198,385,440 bytes against 16,796,017,248. At long context the gap widens: gemma-4-26B-A4B-it's KV cache at 32K is 2.8× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

gemma-4-26B-A4B-itMinistral-3-8B-Reasoning-2512-heretic
Parameters26.5B8.9B
Architecturegemma4mistral3
Layers3034
Native context262,144262,144
Mixture of expertsyes, 128 expertsno
Quantizations published463
Smallest quantization8.99 GiB4.84 GiB
Q4_K_M15.64 GiB4.84 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
Contextgemma-4-26B-A4B-itMinistral-3-8B-Reasoning-2512-hereticRatio
4,0960.45 GiB0.53 GiB1.18×
8,1920.61 GiB1.06 GiB1.75×
16,3840.92 GiB2.13 GiB2.31×
32,7681.54 GiB4.25 GiB2.75×
65,5362.79 GiB8.50 GiB3.04×
131,0725.29 GiB17.00 GiB3.21×