Here are the complete benchmark tables. The model basically dominates in everything evaluated in these tables (Humanity Last Exam, CriPT, ArxivMath, HealthBench, etc.)

→ View original post on X — @dotcsv