New benchmark tests AI models' math skills, revealing significant limitations.

Source: https://arstechnica.com/ai/2024/11/new-secret-math-benchmark-stumps-ai-models-and-phds-alike/