Model comparison

Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled vs Qwen3-Coder-30B-A3B-Instruct

At Q4_K_M, Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled is the smaller download — 16,540,268,960 bytes against 18,556,689,568. At long context the gap widens: Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled'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

Qwen3.5-27B-Claude-4.6-Opus-Reasoning-DistilledQwen3-Coder-30B-A3B-Instruct
Parameters27.8B30.5B
Architectureqwen35qwen3moe
Layers6448
Native context262,144262,144
Mixture of expertsnoyes, 128 experts
Quantizations published3646
Smallest quantization5.80 GiB7.46 GiB
Q4_K_M15.40 GiB17.28 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextQwen3.5-27B-Claude-4.6-Opus-Reasoning-DistilledQwen3-Coder-30B-A3B-InstructRatio
4,0960.25 GiB0.38 GiB1.50×
8,1920.50 GiB0.75 GiB1.50×
16,3841.00 GiB1.50 GiB1.50×
32,7682.00 GiB3.00 GiB1.50×
65,5364.00 GiB6.00 GiB1.50×
131,0728.00 GiB12.00 GiB1.50×