Current AI models solve less than 2% of Frontier Math problems, far below their 90%+ performance on traditional benchmarks.
source
Current AI models solve less than 2% of Frontier Math problems, far below their 90%+ performance on traditional benchmarks.
source