LLM Wins IMO Gold Medal; My First Thought Was 'What Does This Have to Do with Battery Thermal Management?'

Last week, while tuning parameters for a battery thermal management model in the lab, I got stuck on a partial differential equation with strong coupling among multiple variables. Out of habit, I opened my laptop to run a version using conventional numerical methods, only to find the convergence conditions were absurdly strict. A colleague next to me casually said, "Why don't you try that new model from Xiaohongshu (RED)? It just got a perfect score at the IMO." I froze for a second—since when could large models solve problems at this level?