Model comparison

Qwen3-Coder-30B-A3B-Instruct vs Anubis-Mini-8B-v1-heretic

At Q4_K_M, Anubis-Mini-8B-v1-heretic is the smaller download — 4,920,744,992 bytes against 18,556,689,568. At long context the gap widens: Qwen3-Coder-30B-A3B-Instruct's KV cache at 32K is 1.3× 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-InstructAnubis-Mini-8B-v1-heretic
Parameters30.5B8.0B
Architectureqwen3moellama
Layers4832
Native context262,144131,072
Mixture of expertsyes, 128 expertsno
Quantizations published463
Smallest quantization7.46 GiB4.58 GiB
Q4_K_M17.28 GiB4.58 GiB
Licenceapache-2.0llama3.3

KV cache by context

the term that decides long-context viability
ContextQwen3-Coder-30B-A3B-InstructAnubis-Mini-8B-v1-hereticRatio
4,0960.38 GiB0.50 GiB1.33×
8,1920.75 GiB1.00 GiB1.33×
16,3841.50 GiB2.00 GiB1.33×
32,7683.00 GiB4.00 GiB1.33×
65,5366.00 GiB8.00 GiB1.33×
131,07212.00 GiB16.00 GiB1.33×