OpenAI Navier-Stokes researchers say thousands of AI agents helped solve parts of a 90-year-old maths problem in just 88 hours. The claim concerns the Navier-Stokes existence and smoothness problem, one of the famous Millennium Prize problems in mathematics.
### What OpenAI says it achieved
OpenAI says it deployed roughly 10,000 AI agents to work on the problem after an internal model showed strong mathematical abilities. The company says the agents worked continuously for about 88 hours and produced a result covering two of the four statements required for the full Millennium Prize problem.
The Navier-Stokes equations describe how fluids move and are central to the mathematical study of turbulence. Important questions around existence and smoothness have remained unresolved for decades, which is why the problem is considered one of the major challenges in modern mathematics.
OpenAI described the result as a milestone for its AI systems, but it has not presented the claim as a completed Millennium Prize solution. The company also said it does not intend to claim the $1 million prize for the result.
### The result still needs verification
The most important caveat is that the claimed solution has not yet been independently verified or publicly accepted by the Clay Mathematics Institute, which administers the Millennium Prize. A mathematical result of this importance needs detailed checking by independent experts before it can be treated as a confirmed solution.
OpenAI says the agents exchanged nearly 3 million messages and used around 130 billion output tokens while working on Navier-Stokes. Based on OpenAI’s own pricing for advanced model output, the company estimated that an effort of this scale could represent roughly $10 million in compute and model-output costs.
### Why the announcement is controversial
The announcement has also triggered a dispute over the timing and provenance of the work. NYU mathematician Tristan Buckmaster said he and Anthropic mathematician Levent Alpöge had been working on the same problem and raised concerns about whether information about their progress reached OpenAI before its own work was published.
OpenAI said it had not seen their work before it was made public and said no user data was accessed for its Navier-Stokes work. The company also acknowledged that it could not completely rule out whether de-identified data derived from product usage had contributed to model improvements.
For now, the OpenAI Navier-Stokes result is best understood as a significant AI-assisted mathematics claim rather than a verified solution. Independent mathematical review will determine how much of the claimed progress ultimately stands.
