OpenAI is releasing a broad range of new mathematical results produced by an internal frontier model. Before doing so, it consulted the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study and drew on the group’s public recommendations to decide how to publish the work. The results are now in a GitHub repository with defined protocols for paper revisions and citations, and the team says it is still exploring other community-hosted options that meet the committee’s guidelines. OpenAI also commits to improving citations, mathematical exposition, and presentation in future releases.

The repository includes Lean formalizations of many of the proofs, so they can be verified by a computer; more formalizations will be added as they are completed. To support scientific transparency, OpenAI is publishing ten summaries of the model’s reasoning, estimates of compute spent expressed in terms of ChatGPT Pro usage, and statistics about the number of attempted problems. The average result used roughly the equivalent of three hours of ChatGPT Pro thinking.

OpenAI says the goal is to push the frontier of human knowledge and enable further progress in mathematics. It will fund a series of workshops, conferences, and special programs around understanding major AI-produced mathematical results, and it is working to responsibly release the model that produced these results. The company also plans to keep evaluating its internal frontier models on mathematics and other sciences, and says it will update its disclosure standards based on community feedback.