OpenAI announced that it has solved one of the "millennium problems"

OpenAI announced that it has solved one of the "Millennium Prize Problems"

This problem is considered one of the seven "Millennium Prize Problems" announced by the Clay Mathematics Institute at the beginning of the 21st century.

A $1 million prize has been established for solving it.

The Navier-Stokes equations describe the motion of liquids and gases. They can be used to mathematically describe processes such as the flow of oil and gas in pipes, the movement of water, as well as how air or water moves around airplanes, ships, and cars.

The complex aspect of the problem is that scientists have not been able to prove in a general form that in a three-dimensional state, the solution to such equations always exists under any realistic initial conditions and does not "break down" over time.

To put it more simply, the question is: if the initial state of a liquid or gas is known, can the mathematical equation always describe its subsequent motion correctly and continuously? Or can a singularity — a sharp change that is physically meaningless — occur in the mathematical solution in certain situations?

For example, a mathematical model might show that the velocity of a fluid increases infinitely within a certain finite time. However, in the real physical world, such a situation is not observed. For this reason, the existence and smoothness problem of the Navier-Stokes equations is important not only for theoretical mathematics, but also for physics and engineering.

The material states that OpenAI announced it solved this problem on September 8, 2026. Tens of thousands of autonomous AI agents worked on the solution for nearly four days.

During this process, the agents exchanged 4.9 million messages and consumed a total of 300 billion tokens. The author writes that an internal system much more powerful than the models open to OpenAI clients was used for the computation.

Another important aspect of this result is the role of AI in the task. Previously, in reports like "neural network created a new drug" or "AI made a scientific discovery," there were many cases where the main scientific work was performed by human researchers, and artificial intelligence was used as an auxiliary tool at certain stages.

This time, as the material emphasizes, if OpenAI described the process correctly, after the initial task was set, the vast majority of the solution-seeking process was performed by autonomous neural network agents.

In this sense, this result may show that the role of AI in science is changing. That is, the neural network may become not just a "smart calculator" that helps a scientist with calculations, but a tool that performs a massive volume of work in the process of solving an extremely complex problem.