10,000 AI Agents, 88 Hours: OpenAI Solves Navier-Stokes Problem
OpenAI announces 10,000 concurrent AI agents collaborated for 88 hours to produce a complete proof of the Navier-Stokes Millennium Prize Problem, formally verified in Lean.
10,000 AI Agents, 88 Hours: OpenAI Claims Navier-Stokes Millennium Prize Solved
The Second Millennium Problem Falls
On September 8, 2026, OpenAI announced that approximately 10,000 concurrent AI agents collaborated for 88 hours to produce a complete proof of the Navier-Stokes existence and smoothness problem — one of the seven Millennium Prize Problems. The proof, formally verified in Lean, reveals that three-dimensional incompressible fluids can develop singularities in finite time under certain smooth external forces, a phenomenon the team calls "finite-time blowup." This marks the second Millennium Problem claimed solved, following the Poincaré conjecture.
Beyond Any Publicly Available Model
The proof was generated by a next-generation model with capabilities far exceeding the publicly released GPT-6 Astra. The collaboration produced approximately 2.7 million messages and around 130 billion output tokens. This was not a math-specialized system but a general-purpose reasoning model operating at unprecedented scale. The message density across ten thousand agents equates to an uninterrupted academic marathon lasting nearly four days.
What Are the Navier-Stokes Equations?
The Navier-Stokes equations govern fluid motion and underpin modern engineering — from weather forecasting and aircraft design to blood flow and ocean current simulation. Yet for two centuries, mathematicians have been unable to rigorously prove whether solutions in three dimensions remain smooth forever or whether velocities can suddenly "blow up" to infinity at some finite moment. This is precisely what the Clay Mathematics Institute designated as one of its seven Millennium Prize Problems in 2000.
Why This Time May Be Different
Unlike countless previous claims, OpenAI accompanied its announcement with a complete formal verification in Lean. Lean is a language that transforms mathematical proofs into computer-checkable code — if it compiles, every logical step is mechanically sound. While the mathematics community will require months or even years of independent review, formal verification already eliminates the most common source of errors in human proofs: simple oversight.
What Does It Mean?
If ultimately accepted by the mathematical community, this would not only settle a $1 million prize but also mark the first time AI has reached a summit that top human mathematicians could not scale for two centuries. The 10,000-agent collaboration model may itself open a new paradigm for scientific research.