Seven Millennium problems. Several kinds of progress.
Follow the official problem, the proposed advance and the verification still needed.
The Clay Mathematics Institute lists Poincaré among solved problems. Its remaining six problem pages are retained as open targets in this WRS edition. The date and source of that status matter.
OpenAI has released a candidate Navier–Stokes construction with smooth forcing and finite energy, along with a reported Lean formalization. WRS keeps the candidate beside the official problem instead of treating the announcement as an institutional acceptance decision.
The Claude-linked zeta work improves a bound on the proportion of zeros under stated counting conventions. That percentage is not the completion percentage of the Riemann hypothesis. Similarly, formalizing Fermat’s Last Theorem is a verification contribution to an existing theorem, not a new Millennium solution.
Official status checked against Clay Mathematics Institute on . Recent-status notes below describe that check; they are not a complete literature review of ongoing work.
UNSOLVED
P vs NP
Can every efficiently checkable answer also be found efficiently?
Latest status check: The official listing continues to treat the general problem as open; benchmark performance and special algorithms do not settle it.
Can an analytic function reveal the rank of the rational points on an elliptic curve?
Latest status check: The general conjecture remains unresolved in the official listing. Particular curves or conditional results need their own hypotheses.
Do smooth three-dimensional fluid motions always remain well behaved under the problem’s stated conditions?
Latest status check: The full prize problem remains open in the official listing. Simulations, special regimes and partial claims are separate evidence.
Is every closed simply connected three-dimensional manifold a three-dimensional sphere?
Latest status check: The official source identifies Perelman’s proof. This is the solved prize problem; its historical contribution is distinct from new verification exercises.
WRS contributions can inspect a bounded subclaim or proof step. A model-generated candidate, numerical pass or repeated review does not become an accepted solution by appearing here.