OpenAI Model Solves 80-Year-Old Geometry Problem

An internal model developed by OpenAI has disproved a central conjecture in discrete geometry that had remained unresolved for nearly 80 years, in what researchers describe as the first autonomous AI solution to a major open mathematics problem.

The result concerns the planar unit distance problem, first posed by Hungarian mathematician Paul Erdős in 1946. The problem asks: among nnn points placed on a plane, how many pairs can be exactly one unit apart?

For decades, mathematicians believed the maximum number of such pairs grew only slightly faster than linearly, a conjecture associated with Erdős. OpenAI said its model constructed an infinite family of point configurations that exceeded the previously assumed limit by a polynomial factor.

The proof was independently checked by external mathematicians, while a companion paper explaining the argument and its implications was also released. According to OpenAI, the proof emerged from a general-purpose reasoning model rather than a system specifically trained for mathematics.

“This has been one of Erdős’ favourite problems,” said Noga Alon, a combinatorialist at Princeton University. “The solution of the problem by the internal model of OpenAI is, in my opinion, an outstanding achievement, settling a long-standing open problem.”

The problem has been regarded as one of the most important questions in combinatorial geometry because of its simple formulation and resistance to proof. 

Tim Gowers, a Fields Medal winner, described the development as “a milestone in AI mathematics.”

The proof also renewed discussion about AI’s role in scientific research. OpenAI said the result suggests advanced reasoning models may eventually contribute to research in fields including physics, biology, engineering, and materials science by maintaining long chains of reasoning and identifying connections across disciplines.

OpenAI said the work highlights a future in which AI systems assist researchers by exploring technical ideas, while human experts continue to decide which problems matter and how results should be interpreted.