Most state-of-the-art models, scored below 5pct in the 2025 USAMO. One 25pct. LLMs still have a long way to go when it comes to genuine reasoning and #creativity. https://arxiv.org/abs/2503.21934