Model comparison

Qwen3-Coder-30B-A3B-Instruct vs NTX-2.1-Pro

At Q4_K_M, NTX-2.1-Pro is the smaller download — 9,001,754,176 bytes against 18,556,689,568. At long context the gap widens: Qwen3-Coder-30B-A3B-Instruct's KV cache at 32K is 1.7× 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-InstructNTX-2.1-Pro
Parameters30.5B14.8B
Architectureqwen3moeqwen3
Layers4840
Native context262,14440,960
Mixture of expertsyes, 128 expertsno
Quantizations published4635
Smallest quantization7.46 GiB3.33 GiB
Q4_K_M17.28 GiB8.38 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextQwen3-Coder-30B-A3B-InstructNTX-2.1-ProRatio
4,0960.38 GiB0.63 GiB1.67×
8,1920.75 GiB1.25 GiB1.67×
16,3841.50 GiB2.50 GiB1.67×
32,7683.00 GiB5.00 GiB1.67×
65,5366.00 GiB10.00 GiB1.67×
131,07212.00 GiB20.00 GiB1.67×
Qwen3-Coder-30B-A3B-Instruct vs NTX-2.1-Pro — size, memory and hardware fit — ossmodeldb