OpenAI on Sept. 8, 2026, published a claimed resolution of the Navier-Stokes Millennium Prize problem. The company says an analytical proof, plus a Lean formalization, shows that smooth 3D incompressible flow can form a finite-time singularity.
OpenAI says a large internal multi-agent system produced the work.
The same week, New York University's Tristan Buckmaster and Anthropic's Levent Alpoge posted related Euler results. They alleged that OpenAI front-ran their approach after a Sunday call with an OpenAI researcher.
OpenAI's post congratulates the forced-Euler priority and says the proofs differ. After an internal review, the company denies that user Codex data influenced its result.
Terence Tao has warned that agent swarms racing rumors could chill open math sharing.
What remains open is whether the Millennium Prize claim will hold, whether the disputed Sunday call changes the priority fight, and whether any outside researcher data shaped the proof. Company statements and the mathematicians' personal accounts should be weighed separately until those points are settled.