OpenAI Claims AI Solved 90-Year-Old Navier–Stokes Puzzle in 88 Hours

​OpenAI has announced a landmark breakthrough in mathematical research, stating that an unreleased artificial intelligence system utilizing up to 10,000 parallel AI agents has solved a central component of the Navier–Stokes existence and smoothness problem.

​The achievement, completed in approximately 88 hours, targets one of the seven coveted Millennium Prize Problems designated by the Clay Mathematics Institute in 2000, each carrying a $1 million award.

​The Navier–Stokes equations, formulated in the 19th century by Claude-Louis Navier and George Gabriel Stokes, govern the physical mechanics of fluid motion for substances like air and water. For nearly 90 years, mathematicians have wrestled with whether these equations can reliably describe smooth fluid movement indefinitely, or if they inevitably break down through “finite-time blowup”—a theoretical phenomenon where fluid velocities accelerate to infinity within a finite period.

​According to OpenAI, its autonomous agent swarm generated billions of tokens to produce a 165-page analytical proof alongside a machine-checkable formalization written in Lean, a programming language used to verify mathematical logic. The resulting model output indicates that an initially smooth, three-dimensional fluid configuration can indeed develop a singularity. Despite the magnitude of the claim, OpenAI leadership stated that the company does not intend to claim the $1 million prize.

ADVERTISEMENT

​The announcement has immediately sparked intense scrutiny and controversy across the scientific community. Mathematicians Tristan Buckmaster of New York University and Levent Alpöge of Anthropic raised serious concerns regarding the timeline and the potential crossover of ideas, pointing to concurrent independent research they conducted on related fluid dynamics equations.

​OpenAI has strongly defended the independence of its work, maintaining that its internal agents developed the proof without prior access to unpublished human research, while acknowledging that conversations began after rumors of parallel breakthroughs surfaced. As the global mathematics community begins parsing the complex Lean formalizations, the development stands as one of the most polarizing and closely monitored intersections of artificial intelligence and fundamental science to date.