Model comparison

gemma-3-4b-it-qat-q4_0-unquantized vs Qwen2.5-VL-7B-Instruct

At Q4_K_M, gemma-3-4b-it-qat-q4_0-unquantized is the smaller download — 2,489,894,016 bytes against 4,683,072,032.

From the file· summed bytes, KV per layer

Side by side

gemma-3-4b-it-qat-q4_0-unquantizedQwen2.5-VL-7B-Instruct
Parameters4.3B8.3B
Architecturegemma3qwen2vl
Layers28
Native context128,000
Mixture of expertsnono
Quantizations published4350
Smallest quantization1.10 GiB1.93 GiB
Q4_K_M2.32 GiB4.36 GiB
Licencegemmaapache-2.0

KV cache by context

the term that decides long-context viability
Contextgemma-3-4b-it-qat-q4_0-unquantizedQwen2.5-VL-7B-InstructRatio
4,0960.22 GiB
8,1920.44 GiB
16,3840.88 GiB
32,7681.75 GiB
65,5363.50 GiB
131,0727.00 GiB