Model comparison

Qwen3.6-35B-A3B vs Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated

At Q4_K_M, Huihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated is the smaller download — 21,166,753,952 bytes against 21,166,757,728.

From the file· summed bytes, KV per layer

Side by side

Qwen3.6-35B-A3BHuihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliterated
Parameters36.0B36.0B
Architectureqwen35moeqwen35moe
Layers4040
Native context262,144262,144
Mixture of expertsyes, 256 expertsyes, 256 experts
Quantizations published6112
Smallest quantization8.77 GiB12.05 GiB
Q4_K_M19.71 GiB19.71 GiB
Licenceapache-2.0—

KV cache by context

the term that decides long-context viability
ContextQwen3.6-35B-A3BHuihui-Qwen3.5-35B-A3B-Claude-4.6-Opus-abliteratedRatio
4,0960.08 GiB0.08 GiB1.00×
8,1920.16 GiB0.16 GiB1.00×
16,3840.31 GiB0.31 GiB1.00×
32,7680.63 GiB0.63 GiB1.00×
65,5361.25 GiB1.25 GiB1.00×
131,0722.50 GiB2.50 GiB1.00×