OpenAI Releases 722 AI-Generated Math Papers Solving Hundreds of Long-Standing Problems
OpenAI’s October 2026 GitHub release contains 722 manuscripts spanning 372 result families, generated by an unreleased model. AGMAI confirmed solutions to hundreds of open questions; the average result used roughly three hours of ChatGPT Pro thinking time.
OpenAI published a batch of 722 manuscripts in October 2026, presenting solutions to hundreds of long-standing mathematics problems generated by an unreleased AI model. The papers cover 372 result families — groupings of related papers — and were released via a GitHub repository.
The release had been anticipated for weeks. In September 2026, OpenAI said its model had “resolved more than 100 long-standing open problems across most areas of mathematics.” The newly published materials go further, with the independent advisory group AGMAI confirming the batch includes solutions to “hundreds” of open questions. OpenAI says the “average result” required the equivalent of three hours of ChatGPT Pro thinking time, and the release includes summaries of the model’s reasoning, compute estimates, and statistics on the number of problems attempted.
AGMAI — the Advisory Group on Mathematics and Artificial Intelligence — is an independent body of elite mathematicians formed to help communicate AI-generated mathematical results responsibly. In late September, the group published recommendations urging AI labs to release results promptly through established academic channels, disclose technical details such as model names and compute costs, and avoid using mathematical results as marketing tools. The group stated that treating such releases as promotional vehicles “inflicts significant harm on the mathematical community.”
OpenAI stated it plans to improve future releases by refining citations, mathematical exposition, and presentation, and said it is exploring additional community-hosted platforms that meet AGMAI’s guidelines.
The release adds to a growing body of AI-generated mathematical work from OpenAI and rival labs, including results related to a Millennium Prize problem — among the most prominent open questions in mathematics. The pace and handling of these announcements has prompted debate within the mathematical community over research ethics, academic conduct, and how AI companies credit the human mathematicians whose prior work their systems draw upon.