https://arstechnica.com/ai/2024/11/new-secret-math-benchmark-stumps-ai-models-and-phds-alike/
Ars Technica 11/12/2024