LLMs reportedly advance on 4 of 7 Millennium Prize Problems, 3hrs compute each
ycombinator · x · 2026-10-09
Deedy reports OpenAI's math release shows LLMs making substantial progress on 4 of 7 Millennium Prize Problems — Navier-Stokes (claimed), Riemann, Hodge, Birch-Swinnerton-Dyer — each averaging just 3 hours of thinking compute on unreleased models, pending verification. Garry Tan amplified it: "Agent swarms are real."
Related event: Report: OpenAI's Hidden Model Makes Progress on 4 Millennium Prize Problems(2 posts)→
More from AGI Musings
- Frontier AI models beat human experts at earnings predictions for the first time — maithra_raghu · 2026-10-09
- DeepMind VP Nando de Freitas: Corporations Owe a Moral Debt to Academia — NandoDF · 2026-10-09
- NEJM AI editorial examines AI's impact on scientific thinking — zakkohane · 2026-10-09
- Social science learned reproducible repos beat papers — wild math is still far behind — RexDouglass · 2026-10-09
- Ethan Mollick Revisits His 2005 Hacking History Paper to Draw a Parallel With AI — emollick · 2026-10-09
- Thought Experiment: Would Serial-Processing Aliens Change Our Consciousness Judgments? — eschwitz · 2026-10-09