Model comparison

ZAYA1-8B-Coder vs gemma-4-12B-it-qat-q4_0-unquantized

These two publish different quantization sets; the table below has the exact sizes.

From the file· summed bytes, KV per layer

Side by side

ZAYA1-8B-Codergemma-4-12B-it-qat-q4_0-unquantized
Parameters8.8B12.0B
Architecturezayagemma4
Layers8048
Native context131,072262,144
Mixture of expertsyes, 16 expertsno
Quantizations published32
Smallest quantization5.19 GiB6.50 GiB
Q4_K_M5.19 GiB
Licenceapache-2.0

KV cache by context

the term that decides long-context viability
ContextZAYA1-8B-Codergemma-4-12B-it-qat-q4_0-unquantizedRatio
4,0960.31 GiB0.72 GiB2.30×
8,1920.63 GiB0.97 GiB1.55×
16,3841.25 GiB1.47 GiB1.18×
32,7682.50 GiB2.47 GiB1.01×
65,5365.00 GiB4.47 GiB1.12×
131,07210.00 GiB8.47 GiB1.18×