Full Breakdown
OpenAI Announces AI-Generated Proof of Navier-Stokes Millennium Problem Amid Credit Dispute
9/8/2026, 11:25:25 PM
Core Event
On September 8, 2026 OpenAI announced that an unreleased internal model had produced a solution to the Navier–Stokes existence and smoothness problem, one of the seven Clay Mathematics Institute Millennium Prize Problems. The effort required 88 hours of wall-clock time, employed roughly 10,000 parallel AI agents, and generated a formal proof in the Lean theorem-proving system. The claim is presented as a demonstration of the model’s mathematical capability rather than an immediate bid for the $1 million prize.
Background & Context
The Navier–Stokes equations describe fluid motion and underpin weather forecasting, aircraft design, and many engineering simulations. The Millennium Prize formulation asks whether smooth, three-dimensional solutions can develop singularities (“blow-ups”) in finite time. Since the problems were defined in 2000, only the Poincaré conjecture has been officially resolved. Recent AI-driven breakthroughs on other longstanding conjectures have heightened expectations that machine intelligence could tackle the remaining problems.
Timeline
- August 28 – OpenAI began training a new internal model with advanced mathematical capabilities.
- September 1 – The company launched a coordinated effort after hearing rumors that two Millennium problems had been solved elsewhere.
- September 3 – NYU mathematician Tristan Buckmaster reported that information about his and Anthropic researcher Levent Alpöge’s work had reached OpenAI.
- September 8, 2026 – OpenAI publicly announced the Navier–Stokes proof and stated its researchers and agents had not accessed the pair’s unpublished work.
Data & Statistics
- 88 hours of compute produced the proof.
- 10,000 agents worked in parallel; 300 billion output tokens were generated across all problem attempts.
- The compute cost is estimated at $22.5 million.
Official Statements & Responses
OpenAI noted that its proof addresses the forced Navier–Stokes variant, which differs from the unforced formulation posed by the Clay Institute.
Sébastien Bubeck, OpenAI researcher, posted a detailed account denying allegations that the company tried to remove Alpöge from authorship and reiterating that the internal model produced the result with minimal human guidance.
Martin Bridson, president of the Clay Mathematics Institute, said the institute’s evaluation process remains “deliberately unhurried” and will require formal publication and broad expert acceptance before any prize consideration.
Conflicting Reports & Gaps
Sources differ on the exact variant solved: OpenAI’s claim concerns forced Navier–Stokes blow-up, while Buckmaster and Alpöge’s earlier results address the unforced Euler equations. OpenAI asserts the proofs are “significantly different,” yet Buckmaster notes the technical route appears “the same.” Independent mathematicians have not yet examined the full proof, and the Clay Institute still lists the problem as unsolved.
What’s Next
OpenAI’s paper and Lean formalization have been posted for community review. Independent verification will likely involve a multi-month peer-review process, after which the Clay Mathematics Institute may begin its formal evaluation. The dispute over credit may prompt policy discussions within the AI research community regarding data usage, authorship, and the handling of AI-generated mathematical results.
