75
50
8
7
1
4
Target:636
qwen/qwen3.8-flash
((75 + 7) * 8) - (50 - 1)
= 607
Incorrect
openai/gpt-5.6-luna
75 * (8 + 1) - (50 - 7 - 4)
= 636
Correct
z-ai/glm-5.3-flash
(75 * (8 + 1)) - (50 - 7 - 4)
= 636
Correct
meta/muse-spark-1.2-contributor
((75+1)*8)+(7*4)
= 636
Correct
deepseek/deepseek-v4-flash-0731
75*(8+1)-(50-7-4)
= 636
Correct