AI Saturates FrontierMath Benchmarks Once Thought Years Away
FrontierMath Tier 3, once predicted by Terence Tao to resist AI for years, has been fully saturated within two years, with newer GPT models even acing the harder Tier 4. Commenters argue that if even a fraction of this progress rate carried over to biology, pharma and housing, failing to pursue it would be history's biggest anti-progress mistake.
2026-10-03 ~ 2026-10-05 · 3 related posts
- FrontierMath saturated in two years: new GPT solves Tier 4 100%, 9/49 open problems down — Afinetheorem · 2026-10-03
- "The most anti-Progress fumble in human history": extrapolating AI's math pace to biology and energy — Afinetheorem · 2026-10-03
- FrontierMath Tier 3, predicted by Terence Tao to resist AI for years, now saturated — Eliv_nurotic · 2026-10-05