OpenAI revealed solutions to long-standing math problems from an unreleased frontier model in 722 manuscripts across 372 result families, first reported by The Verge. The release arrives amid debate over research ethics, academic conduct and how companies credit human mathematicians whose work their systems build upon and potentially use to produce results.
The 372 result families group related papers. The release also includes some summaries of the model’s reasoning, estimates of compute used and statistics on the number of problems attempted. OpenAI says the average result used the equivalent of three hours of ChatGPT Pro thinking.
In September, OpenAI said its model had resolved more than 100 long-standing open problems across most areas of mathematics. AGMAI says the new release includes solutions to hundreds of open questions.
AGMAI, the Advisory Group on Mathematics and Artificial Intelligence, is a newly formed independent advisory group of elite mathematicians assembled to communicate the results responsibly. It published its first recommendations in late September, urging AI labs to release mathematical results promptly through established academic channels where possible and disclose the model name, prompts and compute costs.
AGMAI also urged AI companies not to treat mathematical-result releases as marketing vehicles for their models, saying that practice inflicts significant harm on the mathematical community.
OpenAI said it is publishing the results in a GitHub repository with protocols for paper revisions and citations. It is continuing to explore other community-hosted alternatives for this release that meet the committee’s guidelines. OpenAI said it is committed, for future releases, to further improving the quality of the papers through citations, mathematical exposition and presentation of the results.
The manuscript counts give the release’s scale, while AGMAI’s recommendations address how results are shared.
The full impact is likely to take time as mathematicians assess and digest the results. They add to a rapidly growing body of mathematical results from OpenAI and rival labs such as Anthropic that the field is still processing, including results concerning a Millennium Prize problem, among the field’s most famous open questions.