OpenAI shares frontier model results on unsolved math problems

OpenAI has published results from an internal frontier model addressing unsolved problems in mathematics. The company shared Lean proof formalizations and detailed research findings on GitHub, making the work accessible to researchers and developers.

The release demonstrates progress in AI reasoning and formal mathematics, a key frontier for AI capability. Sharing the proofs and methodology publicly signals OpenAI's continued focus on advancing the field and enabling others to build on the work.