100
75
25
50
5
4
Target:228
qwen/qwen3.8-flash
50*5 - 100/4 + 75/25
= 228
Correct
z-ai/glm-5.3-flash
4 * ((100 / 50) + (75 + 5 - 25))
= 228
Correct
meta/muse-spark-1.2-contributor
50*5 - 100/4 + 75/25
= 228
Correct
openai/gpt-5.6-luna
((50 - 4) * 5) - ((100 - 75) / 25)
= 229
Incorrect
deepseek/deepseek-v4-flash-0731
4 * (50 + 5 + (100 / (75 - 25)))
= 228
Correct