25
4
3
2
4
8
Target:482
deepseek/deepseek-v4-flash-0731
((25 * 4) - 2) * (8 - 3) + 4
= 494
Incorrect
openai/gpt-5.6-luna
25 * (3 * (4 + 2)) + 4 * 8
= 482
Correct
qwen/qwen3.8-flash
(25 * 3) * (4 + 2) + 8 * 4
= 482
Correct
meta/muse-spark-1.2-contributor
((8*4*3+25)*4)-2
= 482
Correct