OpenAI claims breakthrough in 90-year-old Navier-Stokes math puzzle
An internal AI model solved two parts of a Millennium Prize problem, though the result awaits independent verification.
OpenAI has announced that an internal AI model, described as more capable than GPT-6 Astra, has solved the Navier-Stokes existence and smoothness problem. The breakthrough addresses a mathematical puzzle that has remained unsolved for 90 years.
To reach the solution, OpenAI deployed approximately 10,000 concurrent agents that worked for roughly 88 hours. During the Navier-Stokes phase, these agents exchanged 2.7 million messages and generated approximately 130 billion output tokens. The process began with nearly 100 agents working for about 50 hours on the Euler problem before the larger swarm extended the work to Navier-Stokes. According to OpenAI, the resulting solution resolves statements 'C' and 'D' of the four requirements set by the Millennium Prize formulation. Despite the achievement, the company stated it does not intend to claim the Millennium Prize.
The Fluid Dynamics Puzzle
The Navier-Stokes equations are fundamental to physics, describing the behavior of fluids and gases, such as how air flows over an airplane wing. The Millennium Prize Problem specifically asks whether these equations always produce smooth, predictable solutions or if they can "blow up," reaching a point where they cease to make mathematical sense. This achievement follows a string of recent AI-driven mathematical successes, including the formalization of Fermat's Last Theorem and a solution to a conjecture by Paul Erdős.
A Decoupling of Understanding
If verified, the result suggests that AI can tackle the deepest problems in mathematics, tasks previously thought to require human-level intuition and intricate manual calculations. However, the breakthrough has sparked a debate over the nature of mathematical discovery. UCLA Professor Terence Tao noted that there has been a "strange and unprecedented decoupling" this year between the ability to obtain correct answers and the human understanding of the proofs behind them.
Verification and Controversy
The result has not yet been independently verified by the Clay Mathematics Institute. Beyond the mathematical scrutiny, the announcement is mired in ethical controversy. Researchers Tristan Buckmaster and Levent Alpöge have raised concerns regarding the potential use of their private data via Codex to train the models that achieved the result.
Observers are now waiting for the Clay Mathematics Institute to review the findings. Until then, the industry remains divided on whether this represents a new era of automated discovery or a problematic shift toward "black box" mathematics where the answer exists without a human-comprehensible explanation.