Model comparison

MN-GRAND-23.5B-Gutenberg-UNCENSORED-V2-GLM4.7-Thinking vs gemma-4-12B-it-qat-q4_0-unquantized

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: gemma-4-12B-it-qat-q4_0-unquantized's KV cache at 32K is 4.1× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

MN-GRAND-23.5B-Gutenberg-UNCENSORED-V2-GLM4.7-Thinkinggemma-4-12B-it-qat-q4_0-unquantized
Parameters23.4B12.0B
Architecturellamagemma4
Layers8148
Native context1,024,000262,144
Mixture of expertsnono
Quantizations published332
Smallest quantization5.00 GiB6.50 GiB
Q4_K_M13.37 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextMN-GRAND-23.5B-Gutenberg-UNCENSORED-V2-GLM4.7-Thinkinggemma-4-12B-it-qat-q4_0-unquantizedRatio
4,0961.27 GiB0.72 GiB1.76×
8,1922.53 GiB0.97 GiB2.61×
16,3845.06 GiB1.47 GiB3.45×
32,76810.13 GiB2.47 GiB4.10×
65,53620.25 GiB4.47 GiB4.53×
131,07240.50 GiB8.47 GiB4.78×