Model comparison

MiniCPM5-1B-Agentic-Tooluse-Merged-FP16 vs Qwen3-4B

At Q4_K_M, MiniCPM5-1B-Agentic-Tooluse-Merged-FP16 is the smaller download — 688,065,856 bytes against 2,497,280,256. At long context the gap widens: MiniCPM5-1B-Agentic-Tooluse-Merged-FP16's KV cache at 32K is 6.0× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

MiniCPM5-1B-Agentic-Tooluse-Merged-FP16Qwen3-4B
Parameters1.1B4.0B
Architecturellamaqwen3
Layers2436
Native context131,07240,960
Mixture of expertsnono
Quantizations published629
Smallest quantization0.64 GiB1.01 GiB
Q4_K_M0.64 GiB2.33 GiB
Licenceother

KV cache by context

the term that decides long-context viability
ContextMiniCPM5-1B-Agentic-Tooluse-Merged-FP16Qwen3-4BRatio
4,0960.09 GiB0.56 GiB6.00×
8,1920.19 GiB1.13 GiB6.00×
16,3840.38 GiB2.25 GiB6.00×
32,7680.75 GiB4.50 GiB6.00×
65,5361.50 GiB9.00 GiB6.00×
131,0723.00 GiB18.00 GiB6.00×