Released: 2026-07-15
Key:
Score Threshold:
openai/gpt-4o
openai
((10 + 10) * 50) - 75 - 25 - 100
= 800
mistralai/mistral-medium-3
mistral
(100 - (25 / (10 / 10))) * (75 + 50)
= 9375
deepseek/deepseek-chat-v3-0324
deepseek
(100 + 10) * (75 - (50 - 25)) + 10
= 5510
moonshotai/kimi-k2.5
moonshot
100*10-(75-50/25)
= 927
z-ai/glm-5
z-ai
100 * 10 - 75 + 50 / 25
= 927
google/gemini-3-flash-preview
(10 * (100 - (10 - (25 / 50))))
= 905
anthropic/claude-sonnet-4.6
anthropic
((100 + 75) * 50) / 25 * 10 + 10
= 3510
minimax/minimax-m2.5
minimax
100 * 10 - 75 + 50 / 25
= 927
Each model receives the same prompt with the numbers to use. Models are tasked with creating an expression using only arithmetic operations to reach the target number. Each number can only be used once and you do not have to use all the numbers. Their answers are evaluated without feedback or retries.