Released: 2026-03-23
Key:
Score Threshold:
openai/gpt-4o
openai
(75 * 10) - 100 + 5
= 655
mistralai/mistral-medium-3
mistral
((100 + 50) * (75 - 25)) - (10 * 5)
= 7450
anthropic/claude-sonnet-4.6
anthropic
75 * (100 - 50 - 25) / 5 + 10
= 385
deepseek/deepseek-chat-v3-0324
deepseek
(100 + 25) * 5 + 75 - 10
= 690
minimax/minimax-m2.5
minimax
75 * 10 - (50 + 25 - 5)
= 680
moonshotai/kimi-k2.5
moonshot
75*10 - 50 - 25 + 5
= 680
x-ai/grok-3-mini-beta
x-ai
((100 * (50 / 5)) - (75 + (25 * 10)))
= 675
z-ai/glm-5
z-ai
(75 * 10) - 50 - 25 + 5
= 680
google/gemini-3-flash-preview
75 + 10 * (50 + 100 - 25) / 5
= 325
google/gemini-2.0-flash-001
(100 - 10) * (5 + 75 / 25)
= 720
anthropic/claude-3.7-sonnet
anthropic
(75 + 100 + 5) * 50 / 25
= 360
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.