Model comparison

MaralGPT-Mythos-9B-2606 vs Qwen3.6-27B

At Q4_K_M, MaralGPT-Mythos-9B-2606 is the smaller download — 5,629,109,248 bytes against 16,547,397,888. At long context the gap widens: MaralGPT-Mythos-9B-2606's KV cache at 32K is 2.0× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

MaralGPT-Mythos-9B-2606Qwen3.6-27B
Parameters9.4B27.8B
Architectureqwen35qwen35
Layers3264
Native context1,048,576262,144
Mixture of expertsnono
Quantizations published440
Smallest quantization3.56 GiB8.74 GiB
Q4_K_M5.24 GiB15.41 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextMaralGPT-Mythos-9B-2606Qwen3.6-27BRatio
4,0960.13 GiB0.25 GiB2.00×
8,1920.25 GiB0.50 GiB2.00×
16,3840.50 GiB1.00 GiB2.00×
32,7681.00 GiB2.00 GiB2.00×
65,5362.00 GiB4.00 GiB2.00×
131,0724.00 GiB8.00 GiB2.00×