Model comparison

Noromaid-20b-v0.1.1 vs Qwen3-30B-A3B-Thinking-2507

At Q4_K_M, Noromaid-20b-v0.1.1 is the smaller download — 12,042,208,864 bytes against 18,556,685,824. At long context the gap widens: Qwen3-30B-A3B-Thinking-2507's KV cache at 32K is 12.9× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

Noromaid-20b-v0.1.1Qwen3-30B-A3B-Thinking-2507
Parameters20.0B30.5B
Architecturellamaqwen3moe
Layers6248
Native context4,096262,144
Mixture of expertsnoyes, 128 experts
Quantizations published3351
Smallest quantization4.09 GiB7.05 GiB
Q4_K_M11.22 GiB17.28 GiB
Licencecc-by-nc-4.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextNoromaid-20b-v0.1.1Qwen3-30B-A3B-Thinking-2507Ratio
4,0964.84 GiB0.38 GiB12.92×
8,1929.69 GiB0.75 GiB12.92×
16,38419.38 GiB1.50 GiB12.92×
32,76838.75 GiB3.00 GiB12.92×
65,53677.50 GiB6.00 GiB12.92×
131,072155.00 GiB12.00 GiB12.92×