Model comparison

Qwen3-Coder-30B-A3B-Instruct vs Huihui-Qwen3.6-27B-abliterated

At Q4_K_M, Huihui-Qwen3.6-27B-abliterated is the smaller download — 16,547,399,200 bytes against 18,556,689,568. At long context the gap widens: Huihui-Qwen3.6-27B-abliterated's KV cache at 32K is 1.5× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

Qwen3-Coder-30B-A3B-InstructHuihui-Qwen3.6-27B-abliterated
Parameters30.5B27.8B
Architectureqwen3moeqwen35
Layers4864
Native context262,144262,144
Mixture of expertsyes, 128 expertsno
Quantizations published4617
Smallest quantization7.46 GiB9.98 GiB
Q4_K_M17.28 GiB15.41 GiB
Licenceapache-2.0agpl-3.0

KV cache by context

the term that decides long-context viability
ContextQwen3-Coder-30B-A3B-InstructHuihui-Qwen3.6-27B-abliteratedRatio
4,0960.38 GiB0.25 GiB1.50×
8,1920.75 GiB0.50 GiB1.50×
16,3841.50 GiB1.00 GiB1.50×
32,7683.00 GiB2.00 GiB1.50×
65,5366.00 GiB4.00 GiB1.50×
131,07212.00 GiB8.00 GiB1.50×