OpenAI says an unreleased internal model — stronger than GPT-6 Astra—produced a proof that 3D Navier–Stokes flows can form a finite-time singularity, a Millennium Prize problem. The claim is big, but it remains a proposed solution until independent mathematicians verify the proof.
The method was not a single chatbot answer. OpenAI says about 10,000 AI agents worked in parallel for roughly 88 hours, then used GPT-6 Astra to formalize and check the result in the Lean proof assistant. If correct, this shows AI can act as a research engine that generates and verifies difficult new math.
A dispute has erupted over credit. NYU’s Tristan Buckmaster and Anthropic’s Levent Alpöge had been pursuing similar blow-up results and used AI tools, including OpenAI’s Codex. Buckmaster questions the timeline and possible influence of his drafts; OpenAI denies seeing their work or accessing specific user data.
The key question now is independent verification and fair credit. A proof assistant can check that steps follow rules, but experts must still judge assumptions, novelty, and whether the proof exactly matches the Millennium problem.
THE REAL TEST IS PUBLIC PROOF, INDEPENDENT VALIDATION, AND FAIR CREDIT.
Sanjay Sahay
Have a nice evening.