Model comparison

WizardLM-Uncensored-SuperCOT-StoryTelling-30b vs Qwen3-Coder-30B-A3B-Instruct

At Q4_K_M, Qwen3-Coder-30B-A3B-Instruct is the smaller download — 18,556,689,568 bytes against 19,621,149,152. At long context the gap widens: Qwen3-Coder-30B-A3B-Instruct's KV cache at 32K is 16.3× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

WizardLM-Uncensored-SuperCOT-StoryTelling-30bQwen3-Coder-30B-A3B-Instruct
Parameters32.5B30.5B
Architecturellamaqwen3moe
Layers6048
Native context2,048262,144
Mixture of expertsnoyes, 128 experts
Quantizations published1246
Smallest quantization12.58 GiB7.46 GiB
Q4_K_M18.27 GiB17.28 GiB
Licenceotherapache-2.0

KV cache by context

the term that decides long-context viability
ContextWizardLM-Uncensored-SuperCOT-StoryTelling-30bQwen3-Coder-30B-A3B-InstructRatio
4,0966.09 GiB0.38 GiB16.25×
8,19212.19 GiB0.75 GiB16.25×
16,38424.38 GiB1.50 GiB16.25×
32,76848.75 GiB3.00 GiB16.25×
65,53697.50 GiB6.00 GiB16.25×
131,072195.00 GiB12.00 GiB16.25×