OpenAI says its AI agents have produced a proof of one of mathematics’ biggest unsolved problems. The claim has sparked a sharp dispute over who deserves the credit.

The problem is the Navier-Stokes equations, which describe fluid flow. It is one of the Clay Mathematics Institute’s seven Millennium Prize Problems, each worth 1 million dollars.

OpenAI says it deployed about 10,000 agents on an internal model. After 88 hours starting September 1, the agents produced a 166-page paper with a computer-verified Lean formalisation. The agents exchanged millions of messages and generated hundreds of billions of tokens.

But hours before the announcement, mathematicians Tristan Buckmaster of NYU and Levent Alpoge of Anthropic posted their own results. They had spent most of a year using OpenAI Codex and Anthropic Claude to chase a related proof.

Buckmaster suggested that OpenAI’s agents may have benefited from work he and Alpoge did using OpenAI’s models. OpenAI has denied that any agent or employee accessed their work.

Terence Tao called the result a remarkable achievement. The Clay Institute requires publication, a two-year wait and general acceptance before considering a solution. OpenAI says it is not seeking the prize.