Model comparison

Parable-Granite-4.1-3B-Claude-Fable-5 vs Qwen3-4B

At Q4_K_M, Parable-Granite-4.1-3B-Claude-Fable-5 is the smaller download — 2,099,503,232 bytes against 2,497,280,256. At long context the gap widens: Parable-Granite-4.1-3B-Claude-Fable-5's KV cache at 32K is 1.8× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

Parable-Granite-4.1-3B-Claude-Fable-5Qwen3-4B
Parameters3.4B4.0B
Architecturegraniteqwen3
Layers4036
Native context131,07240,960
Mixture of expertsnono
Quantizations published3629
Smallest quantization0.79 GiB1.01 GiB
Q4_K_M1.96 GiB2.33 GiB
Licenceapache-2.0

KV cache by context

the term that decides long-context viability
ContextParable-Granite-4.1-3B-Claude-Fable-5Qwen3-4BRatio
4,0960.31 GiB0.56 GiB1.80×
8,1920.63 GiB1.13 GiB1.80×
16,3841.25 GiB2.25 GiB1.80×
32,7682.50 GiB4.50 GiB1.80×
65,5365.00 GiB9.00 GiB1.80×
131,07210.00 GiB18.00 GiB1.80×