On October 6, 2026, OpenAI put 722 manuscripts from an unreleased internal model on GitHub, including a zero-free region for the Riemann zeta function and the rational Hodge conjecture for CM abelian varieties. None is independently confirmed, Lean documentation covers only 235 of 372 result families, and some mathematicians are asking for proof.
Also in news.google.com, officechai.com, americantalk.live, livescience.comMillennium Prize Problems
Real progress on the six open Millennium Prize Problems — the Riemann hypothesis, P vs NP, Navier–Stokes, Yang–Mills, Hodge and Birch–Swinnerton-Dyer: serious papers, partial results, AI-assisted advances, and what experts say about claimed solutions.
Latest
Oct 7 · 3 of 13 shown- Claims include a Riemann zero-free region, rational Hodge for CM abelian varieties, and conditional BSD leading term. Experts have not confirmed them.
- The Navier–Stokes blow-up built by OpenAI agents needs external forcing, so the unforced problem stays open, Klainerman says.
- Mathematicians say blog-style releases hinder verification. Lean covers only 235 of 372 result families.
Among the 722 manuscripts in 372 result families is a claimed proof of the full Birch–Swinnerton-Dyer leading-term formula for elliptic curves over the rationals, under conditions on the corank of a q-power Selmer group. This is narrower than the full conjecture, and specialists have yet to check it independently.
Also in quantamagazine.org, news.google.com, officechai.com, newscientist.comRongchang Liu, Kening Lu and Lin Shi prove a local Hartman–Grobman theorem at zero for the unforced 3D Navier–Stokes equation on the torus, conjugating its flow to the heat semigroup in H^s for s>7/2. It is a local result near zero and does not address the regularity or blow-up question itself.
More in Science
Streams that already watch this field. Follow one as it is — it costs nothing extra.
Where the P vs NP question stands: circuit and proof-complexity lower bounds, barriers, meta-complexity, geometric complexity theory, new reductions and hardness results, and how the community reads each claimed proof.
A regular roundup of the latest science news explained in plain language for readers without a science background.
Progress in quantum hardware and error correction — experiments on real devices, not just proposals.