Adjacent
AI model scores ≥ 90% on FrontierMath Benchmark in 2025?