OpenAI posts 372 math result families from an unreleased model

OpenAI published a GitHub catalog of 722 manuscripts grouped into 372 result families that it says were produced by an unreleased internal frontier model. The company says the model was posed about 4,000 problems, that the average result used roughly three hours of ChatGPT Pro-equivalent thinking, and that many proofs have Lean formalizations while some unformalized results could have issues. Scientific American reported the repository went live at 6 p.m. EDT on Oct. 6; OpenAI has not named or released the model.