OpenAI releases a broad set of new math results from an internal frontier model

OpenAI says it is releasing a broad range of new mathematical results produced by an internal frontier model, published at github.com/openai/math.

Greg Brockman frames the release as aimed at accelerating scientific discovery and improving quality of life, and says the model was developed with advice from the Institute for Advanced Study's Advisory Group on Mathematics and Artificial Intelligence.

Andrew Curran says the unnamed internal model, which he calls Aeon, solved most of the recent results in one shot from a single prompt, with about three hours of thinking per result across 4,000 problems OpenAI gave it. He says the model did not exist before the end of August and is still training. These are Curran's claims, not confirmed by OpenAI in the reports reviewed, and the reports do not establish how the results were verified or peer-reviewed.