OpenAI publishes 722 AI-generated math manuscripts on GitHub

OpenAI has posted 722 mathematical manuscripts organized into 372 families and classified by discipline, produced by an unreleased internal model that was posed roughly 4,000 problems and spent, on average, three hours of ChatGPT Pro thinking compute per result.

The README is explicit that verification varies across the catalog. 'Not all have accompanying Lean formalizations,' it reads, adding bluntly: 'Some of the unformalized results could have issues. We will endeavor to fix any such issues quickly.' OpenAI frames the project as a byproduct of model evaluation, writing that it expanded this effort 'after performance on our existing mathematical evaluations saturated.'

Alongside the preprints, OpenAI released abridged reasoning summaries for ten families, from the irrationality exponent of π to the three-dimensional relativistic Vlasov–Maxwell system. Two results sit outside the standard procedure: work on 'a zero-free region for the Riemann zeta function and proof of the Hodge Conjecture for CM abelian varieties.' The writeup for the Re(s) > 11/12 zero-free region was human-edited for readability. Nine researchers from the AI Weekly Who's Who directory passed the link around the day it dropped, treating it as a flagship release rather than another model-development artifact.