An AI Model May Have Cracked One of Mathematics' Hardest Problems
On 8 September 2026 OpenAI said an unreleased AI model proved that the 3D Navier–Stokes equations can blow up in finite time, backed by a 166-page paper and a Lean formal proof. Mathematicians are still checking it, and no Millennium Prize has been awarded.
By the UISC BD Editorial Desk · United Information Service Center · Published 13 September 2026 · 6-minute read
"OpenAI Navier–Stokes" became one of the fastest-rising AI searches on Google this week. The reason: a claim that artificial intelligence has settled a question mathematicians have worked on for generations.
What the Navier–Stokes Equations Are
The Navier–Stokes equations describe how fluids move — water in a pipe, air over a wing, weather systems, blood in arteries. Engineers use them every day.
Yet mathematicians have never proved a basic fact about them in three dimensions: whether smooth, well-behaved solutions always stay smooth, or whether they can "blow up" — develop a singularity — in a finite time. The Clay Mathematics Institute lists this as one of its seven Millennium Prize Problems, each with a $1 million award.
What OpenAI Claims
- On 8 September 2026, OpenAI announced that an unreleased internal model proved that a finite-time singularity occurs in the three-dimensional incompressible Navier–Stokes equations.
- It published a 166-page paper with public proof files.
- It says the proof was formally verified using Lean, an interactive theorem prover in which every logical step is checked by software.
- The effort reportedly used about $22 million of computing power over six days, with 10,000 AI agents.
Why Formal Verification Matters
A traditional proof of this length could take experts months or years to check. A Lean proof is different: if the statement is formalised correctly, the software confirms each step follows from the last.
The key caveat is that phrase "formalised correctly". Mathematicians still need to confirm that the formal statement matches the real problem, with no hidden assumptions.
The Dispute Over Credit
Just before OpenAI's announcement, Tristan Buckmaster, a mathematician at New York University's Courant Institute, released a statement saying that he and Levent Alpöge, a mathematician working at Anthropic, had used AI models from both Anthropic and OpenAI to find solutions to related problems — including a strikingly similar result for part of the Navier–Stokes question.
That has turned a scientific milestone into a debate about who got there first and how AI-assisted discoveries should be credited.
Where Things Stand
A proof artefact is public and can be scrutinised. But broad independent verification, formal publication and recognition by the Clay Mathematics Institute have not happened yet. The accurate description today is a machine-verified proof claim, not a settled Millennium Problem.
Why It Matters Either Way
Even if details are revised, the episode shows AI systems moving from answering questions to producing original, checkable research at the frontier of mathematics. Combined with formal verification, that could speed up discovery in physics, engineering and beyond.
For the model behind the headlines, see GPT-6 Astra explained. For why AI output still needs checking outside formal proofs, see why AI makes things up.
Related reading
- Tesla Robotaxi Passes 1 Million Unsupervised Miles: What It Means
- GPT-6 Astra Explained: What OpenAI's New Model Can Do and What It Costs
- iPhone Duo: Price, Specs and Release Date of Apple's First Foldable
- iPhone 18 Pro and Pro Max: What's New, Price and Should You Upgrade
Sources
- "OpenAI says it cracked Navier-Stokes, one of math's grand challenges," Fortune — fortune.com
- "OpenAI claims blockbuster math breakthrough amid swirl of controversy," Scientific American — scientificamerican.com
- "OpenAI's claimed Navier-Stokes proof raises the ceiling for AI research," The Rundown AI — therundown.ai
- "OpenAI's Navier–Stokes claim awaits independent review," Quasa — quasa.io
