OpenAI Published a Proposed Solution to the Navier–Stokes Millennium Prize Problem

OpenAI Published a Proposed Solution to the Navier–Stokes Millennium Prize Problem

• AI & Machine Learning
• September 09, 2026

?OpenAI has published what it describes as a solution to the Navier–Stokes existence and smoothness problem, one of the seven Millennium Prize Problems. According to the company, its internal AI system produced an analytical proof and a formalization in Lean showing that an initially smooth fluid at rest, under a smooth applied force, can develop a singularity in finite time while its energy remains finite. The problem has remained unresolved for roughly 90 years and concerns equations used to model fluid motion in areas such as aircraft design, weather forecasting, and blood flow.

What makes the announcement particularly interesting from an AI engineering perspective is how the result was produced. The proof was not generated through a single model interaction or one long reasoning chain. OpenAI used a coordinated system of agents powered by its internal model, with different groups exploring different formulations and approaches in parallel. The group that produced the Navier–Stokes result involved around 10,000 concurrent agents, while Codex was used to consolidate useful intermediate findings across groups.

The scale was substantial. OpenAI reports that the agents arrived at the result roughly 88 hours after the effort began, followed by another 17 hours of Lean formalization and verification. The Navier–Stokes effort alone involved 2.7 million agent messages and approximately 130 billion output tokens. Rather than demonstrating only stronger model reasoning, the experiment shows how AI-driven research can be organized as large-scale parallel exploration, with multiple agent groups pursuing different approaches, sharing intermediate findings, and feeding promising results into subsequent work.

From an engineering perspective, this may be the more interesting part of the announcement. The experiment suggests that progress in scientific AI may depend not only on stronger individual models, but also on the systems built around them: how research tasks are distributed, how agents coordinate, how intermediate knowledge is consolidated, and how generated results are formally verified.

There is still an important distinction between OpenAI publishing a proposed solution and the mathematical community accepting the Millennium Prize Problem as resolved. OpenAI says it does not intend to claim the Millennium Prize. Regardless of the eventual assessment of this particular proof, the experiment provides an important example of another direction for AI-assisted research: coordinated systems that can explore many reasoning paths in parallel and then subject the resulting work to formal verification.

Source: OpenAI, “On the Navier–Stokes Millennium Prize Problem,” September 8, 2026.

Image credit: Luke Jones / Unsplash