OpenAI shares new math frontier model results, GitHub release with Princeton IAS advisory group
OpenAI released a GitHub repo of math frontier results on October 6, partnering with Princeton IAS's AGMAI advisory group, including Lean formalizations, with each result using about 3 hours of ChatGPT Pro compute on average.
On October 6, 2026, OpenAI announced new math research progress, releasing a GitHub repository containing several new math results produced by an internal frontier model. In this project, OpenAI collaborated with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study (AGMAI), following their public recommendations to improve how results are shared.
For this release, OpenAI used a GitHub repository as the publication platform, with protocols for paper revisions and citations. To promote scientific transparency and openness, the repo also shares additional details about how the results were obtained, including 10 summaries of the model's reasoning, compute spent estimated in terms of ChatGPT Pro usage, and statistics on attempted problems. On average, each result used compute equivalent to about 3 hours of ChatGPT Pro thinking.
On formalization, OpenAI is formalizing many proofs in the Lean programming language, allowing mathematical proofs to be checked by a computer. These formalizations will be continuously updated to the repository as more results are obtained. OpenAI says future releases will continue to improve paper quality through citations, mathematical exposition, and presentation of results.
On research community building, OpenAI says it will fund a series of workshops, conferences, and special programs around understanding major math results produced by AI, with more details to follow. OpenAI also plans to responsibly release the internal frontier model that produced these results, and to continue evaluating the model on mathematics and other sciences.
The AGMAI advisory group is operated by the Princeton Institute for Advanced Study, focused on providing methodological guidance for AI and math research. OpenAI's three-layer transparency mechanism of GitHub repo + Lean formalization + compute disclosure, distinct from the traditional "PDF paper + closed model" release model, is seen by the industry as a transformation in the way math AI research is published.