Model comparison

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

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

gemma-4-12B-it-qat-q4_0-unquantizedMN-GRAND-23.5B-Gutenberg-UNCENSORED-V2-GLM4.7-Thinking
Parameters12.0B23.4B
Architecturegemma4llama
Layers4881
Native context262,1441,024,000
Mixture of expertsnono
Quantizations published233
Smallest quantization6.50 GiB5.00 GiB
Q4_K_M13.37 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
Contextgemma-4-12B-it-qat-q4_0-unquantizedMN-GRAND-23.5B-Gutenberg-UNCENSORED-V2-GLM4.7-ThinkingRatio
4,0960.72 GiB1.27 GiB1.76×
8,1920.97 GiB2.53 GiB2.61×
16,3841.47 GiB5.06 GiB3.45×
32,7682.47 GiB10.13 GiB4.10×
65,5364.47 GiB20.25 GiB4.53×
131,0728.47 GiB40.50 GiB4.78×
gemma-4-12B-it-qat-q4_0-unquantized vs MN-GRAND-23.5B-Gutenberg-UNCENSORED-V2-GLM4.7-Thinking — size, memory and hardware fit — ossmodeldb