Model comparison

Qwen3-Coder-30B-A3B-Instruct vs gpt-oss-20b

At Q4_K_M, gpt-oss-20b is the smaller download — 11,624,759,488 bytes against 18,556,689,568. At long context the gap widens: gpt-oss-20b's KV cache at 32K is 3.9× 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-Instructgpt-oss-20b
Parameters30.5B21.5B
Architectureqwen3moegpt-oss
Layers4824
Native context262,144131,072
Mixture of expertsyes, 128 expertsyes, 32 experts
Quantizations published4615
Smallest quantization7.46 GiB10.68 GiB
Q4_K_M17.28 GiB10.83 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextQwen3-Coder-30B-A3B-Instructgpt-oss-20bRatio
4,0960.38 GiB0.11 GiB3.37×
8,1920.75 GiB0.21 GiB3.66×
16,3841.50 GiB0.39 GiB3.82×
32,7683.00 GiB0.77 GiB3.91×
65,5366.00 GiB1.52 GiB3.95×
131,07212.00 GiB3.02 GiB3.98×