OpenAI says an unreleased next-generation artificial-intelligence model has produced a solution to the Navier–Stokes existence-and-smoothness problem, one of the seven famed Millennium Prize Problems established in 2000. The company says approximately 10,000 coordinating AI agents worked for 88 hours to construct a proof demonstrating that smooth three-dimensional fluid motion governed by the Navier–Stokes equations can develop a singularity in finite time. The result includes both an analytical proof and a formalized version in the Lean theorem prover, allowing the mathematical logic to be mechanically checked. If the proof survives independent examination, it would mark only the second Millennium Prize Problem to be solved and arguably the most consequential demonstration yet that artificial intelligence can independently produce major new mathematical knowledge. OpenAI says it does not intend to claim the associated $1 million prize, while mathematicians are now beginning the essential process of scrutinizing the result.
Key Takeaways
- OpenAI says its unreleased AI system solved the roughly 90-year-old Navier–Stokes existence-and-smoothness problem in approximately 88 hours using around 10,000 coordinating AI agents, producing a proof that fluid equations can develop a singularity in finite time.
- The claimed solution has been formalized using Lean, providing machine verification of its logical structure, but independent mathematicians must still examine the assumptions, construction and implications before the result can be regarded as conclusively established.
- If confirmed, the accomplishment could represent a watershed for artificial intelligence: a transition from AI primarily assisting human mathematical research to AI systems independently discovering solutions to problems that have resisted generations of leading mathematicians.
In-Depth
OpenAI’s claim that an unreleased artificial-intelligence system solved the Navier–Stokes existence-and-smoothness problem represents a potentially historic advance, but the proper response is verification rather than celebration by press release. The problem, one of seven Millennium Prize Problems established in 2000, asks whether equations describing three-dimensional fluid motion can develop singularities—points where smooth behavior breaks down. OpenAI says its system produced a proof showing such a breakdown can occur in finite time.
The effort itself is striking. OpenAI says roughly 10,000 coordinating AI agents worked for about 88 hours, producing an analytical argument and a formalized proof using Lean, a theorem-proving system capable of mechanically checking logical steps. If independent mathematicians validate the result, artificial intelligence will have crossed an important threshold: from helping researchers manipulate calculations and explore conjectures to independently attacking one of mathematics’ most celebrated unresolved questions.
Yet institutional scientific caution remains essential. A machine-checked proof is not automatically equivalent to broad mathematical acceptance, particularly when the formulation, assumptions and interpretation must still withstand expert scrutiny. The Millennium Prize process was designed precisely to prevent extraordinary claims from becoming established facts through publicity alone.
There is also a larger consequence. Mathematics provides an unusually clean test of intelligence because answers cannot be rescued by rhetoric, polling or subjective interpretation. If AI systems can reliably discover genuinely new proofs at this level, their significance extends beyond mathematics. Such systems could accelerate physics, engineering and other technical disciplines. The prudent course is neither dismissal nor hype, but rigorous independent examination.
Sources
- https://www.nature.com/articles/d41586-026-02842-5
- https://www.washingtonpost.com/technology/2026/09/09/openai-claims-it-solved-elusive-math-problem-with-1-million-prize/
- https://www.axios.com/2026/09/08/openai-math-solution-navier-stokes-credit
- https://www.wsj.com/tech/ai/openai-millennium-prize-navier-stokes-math-2bf240f8