OpenAI announced today that one of its reasoning models has solved the planar unit distance problem, an 80-year-old mathematical conjecture that has puzzled mathematicians since Paul Erdős first posed it in 1946. The breakthrough marks the first time artificial intelligence has autonomously resolved a prominent open problem central to a field of mathematics.
The unit distance problem asks a deceptively simple question: if you place n points in a plane, how many pairs of points can be exactly distance 1 apart? For nearly eight decades, mathematicians believed that square grid constructions were essentially optimal for maximizing unit-distance pairs, with the prevailing conjecture suggesting an upper bound of n^(1+o(1)).
OpenAI's model has disproved this longstanding belief, discovering "an infinite family of examples that yield a polynomial improvement" over the square grid approach. The proof uses sophisticated techniques from algebraic number theory applied to an elementary geometric question—an unexpected connection that impressed leading mathematicians.
Redemption After Earlier Misstep
This announcement comes seven months after OpenAI faced embarrassment when former VP Kevin Weil claimed GPT-5 had solved 10 previously unsolved Erdős problems. That claim was quickly debunked—the model had simply found solutions that already existed in the literature. Rivals including Yann LeCun and Google DeepMind CEO Demis Hassabis mocked the premature announcement, forcing Weil to delete his post.
This time, OpenAI took a different approach. The company published companion remarks from respected mathematicians who verified the proof, including Noga Alon, Melanie Wood, and Thomas Bloom, who maintains the Erdős Problems website and had previously criticized Weil's "dramatic misrepresentation."
"There is no doubt that the solution to the unit-distance problem is a milestone in AI mathematics," said Fields medalist Tim Gowers. "If a human had written the paper and submitted it to the Annals of Mathematics and I had been asked for a quick opinion, I would have recommended acceptance without any hesitation."
Beyond Mathematical Assistance
What makes this achievement particularly notable is how it was discovered. The proof came from a general-purpose reasoning model rather than a system specifically trained for mathematics or targeted at this particular problem. As part of evaluating whether advanced models can contribute to frontier research, OpenAI tested the system on a collection of Erdős problems.
"In my opinion this paper demonstrates that current AI models go beyond just helpers to human mathematicians—they are capable of having original ingenious ideas, and then carrying them out to fruition," said leading number theorist Arul Shankar.
The model's chain of thought reveals interesting patterns. According to Shankar, "a significant majority of the thoughts are trying to construct a counterexample to the widely believed upper bound, rather than trying to prove it. This argues that the model has some combination of good intuition, willingness to try approaches considered long-shot by the community, and a predisposition to attempt constructions."
Implications for Scientific Research
The breakthrough demonstrates AI's growing capability to handle complex reasoning chains and connect ideas across mathematical fields. OpenAI also revealed that GPT-5.2, their latest model, has been contributing to scientific work across mathematics, physics, biology, and computer science.
In another recent case study, GPT-5.2 Pro helped resolve an open research problem in statistical learning theory, documented in a new paper on learning-curve monotonicity for maximum likelihood estimators. The model solved the problem directly without human-provided proof strategies or intermediate arguments.
On the GPQA Diamond benchmark, which tests graduate-level scientific knowledge, GPT-5.2 Pro achieved 93.2%. On FrontierMath, an evaluation of expert-level mathematics, GPT-5.2 Thinking solved 40.3% of problems—setting a new state of the art.
Noga Alon, a leading combinatorialist at Princeton who called the unit distance problem "one of Erdős' favorite problems," emphasized its significance: "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 proof is available in OpenAI's published papers, along with companion remarks from external mathematicians and an abridged version of the model's reasoning process. As mathematician Thomas Bloom noted, "AI is helping us to more fully explore the cathedral of mathematics we have built over the centuries. What other unseen wonders are waiting in the wings?"
