OpenAI Claims Its AI System Solved a Millennium Math Problem

The Navier–Stokes equations describe the behavior of fluids, including water and air, and are among the seven Millennium Prize Problems established in 2000 by the Clay Mathematics Institute.
A solution to the Navier–Stokes problem would qualify for a $1 million Millennium Prize; before OpenAI’s claim, only one of the seven problems had been solved.
OpenAI researcher Sébastien Bubeck said the company began training a new mathematical-reasoning model on Aug. 28 and committed additional resources after hearing rumors that Anthropic researchers were making progress on the problem.
Describing the moment the system produced its result, Bubeck said: “I thought there had to be a mistake somewhere. And on Sunday morning, we had the final solution, Lean formalized and all.”
The announcement has prompted both excitement and concern among mathematicians, as AI systems increasingly tackle problems that had long been viewed as a pinnacle of human mathematical achievement.
OpenAI claims its artificial intelligence has solved the Navier–Stokes equations, one of the seven hardest unsolved problems in mathematics. The Verge reports the company used over 1,000 AI agents working for more than 50 hours to crack the 90-year-old puzzle, formally verifying the solution in the Lean programming language. The breakthrough would make it only the second Millennium Prize Problem ever solved and potentially earn OpenAI a $1 million prize from the Clay Mathematics Institute.
But the announcement has sparked fierce debate. France24 notes that mathematician Tristan Buckmaster and other researchers have challenged OpenAI's account, claiming the company rushed to publish after learning competitors were close to a solution. The proof requires independent verification from the mathematics community before the claim can be considered legitimate.
The Navier–Stokes equations describe how fluids move. Water flowing down a drain, air currents around an airplane wing, smoke drifting through a room — all follow these mathematical rules. BNN Bloomberg explains that scientists have used these equations for nearly two centuries but could never prove they always work.
The Clay Mathematics Institute listed it as one of seven Millennium Prize Problems in 2000. Solving it would be worth $1 million and represent one of humanity's greatest mathematical achievements. Before OpenAI's claim, only one of these seven problems had ever been solved.
OpenAI researcher Sébastien Bubeck revealed the timeline: the company started training a new math-focused AI model on August 28. The Verge reports that when Bubeck heard rumors Anthropic researchers were close to solving Navier–Stokes, OpenAI doubled down, investing millions more in computing power.
The breakthrough came quickly. Bubeck described the moment: "I thought there had to be a mistake somewhere. And on Sunday morning, we had the final solution, Lean formalized and all." The entire effort consumed massive computational resources but took only days to reach the answer.
Tristan Buckmaster and other researchers say OpenAI's story doesn't add up. The Verge reports they allege the company accelerated its work specifically after learning about their progress on the same problem. They also claim OpenAI tried to shape how credit would be assigned for the discovery.
No independent mathematician has yet verified the proof. France24 notes the mathematics community is divided—some excited AI can tackle such problems, others concerned about whether the result represents genuine human understanding or just computational brute force. The verification process could take weeks or months.
Publishers
85
Articles
134
Reach
219