Model comparison

Qwen3-Coder-30B-A3B-Instruct vs Ling-mini-2.0

At Q4_K_M, Ling-mini-2.0 is the smaller download — 9,911,575,072 bytes against 18,556,689,568. At long context the gap widens: Ling-mini-2.0's KV cache at 32K is 2.4× 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-InstructLing-mini-2.0
Parameters30.5B16.3B
Architectureqwen3moebailingmoe2
Layers4820
Native context262,14432,768
Mixture of expertsyes, 128 expertsyes, 256 experts
Quantizations published4630
Smallest quantization7.46 GiB4.34 GiB
Q4_K_M17.28 GiB9.23 GiB
Licenceapache-2.0—

KV cache by context

the term that decides long-context viability
ContextQwen3-Coder-30B-A3B-InstructLing-mini-2.0Ratio
4,0960.38 GiB0.16 GiB2.40×
8,1920.75 GiB0.31 GiB2.40×
16,3841.50 GiB0.63 GiB2.40×
32,7683.00 GiB1.25 GiB2.40×
65,5366.00 GiB2.50 GiB2.40×
131,07212.00 GiB5.00 GiB2.40×