Evaluation of AI potential for solving higher mathematics olympiad challenges

Abstract

Problems from mathematical olympiads, renowned for their high complexity and the necessity for a deep understanding of mathematical concepts, represent an intriguing direction for exploring the capabilities of AI models. Solving olympiad problems poses a significant challenge for contemporary AI models. The results of testing various AI models on problems from established olympiads such as AMC, AIME, IMO, and other sources are considered in this study.