Model comparison

EXAONE-4.0-32B vs Qwen3-Coder-30B-A3B-Instruct

At Q4_K_M, Qwen3-Coder-30B-A3B-Instruct is the smaller download — 18,556,689,568 bytes against 19,343,826,720.

From the file· summed bytes, KV per layer

Side by side

EXAONE-4.0-32BQwen3-Coder-30B-A3B-Instruct
Parameters32.0B30.5B
Architectureexaone4qwen3moe
Layers6448
Native context131,072262,144
Mixture of expertsnoyes, 128 experts
Quantizations published3146
Smallest quantization8.96 GiB7.46 GiB
Q4_K_M18.02 GiB17.28 GiB
Licenceotherapache-2.0

KV cache by context

the term that decides long-context viability
ContextEXAONE-4.0-32BQwen3-Coder-30B-A3B-InstructRatio
4,0961.00 GiB0.38 GiB2.67×
8,1921.34 GiB0.75 GiB1.79×
16,3841.84 GiB1.50 GiB1.23×
32,7682.84 GiB3.00 GiB1.05×
65,5364.84 GiB6.00 GiB1.24×
131,0728.84 GiB12.00 GiB1.36×
EXAONE-4.0-32B vs Qwen3-Coder-30B-A3B-Instruct — size, memory and hardware fit — ossmodeldb