OpenAI math reached a new stage on October 6, 2026: the company published a broad set of results produced by an internal frontier model that is not yet available to the public. The material went to the openai/math GitHub repository, with revision and citation protocols.
According to the company, the release follows consultation with the Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study and the group's public recommendations. OpenAI says future disclosures will improve citations, mathematical exposition, and presentation.
What is in the repository
The README describes the current catalog: 722 manuscripts organized into 372 families. A family groups a principal result, companion arguments, consequences, or alternative proofs. Not every paper has a Lean formalization. The company says it will update the repository as more formalizations are ready and acknowledges that some unformalized results may have issues.
The package also includes ten abridged summaries of the model's reasoning. Listed topics include the irrationality exponent of π, the Mahler conjectures, Kaplansky's direct-finiteness conjecture in characteristic two, and the three-dimensional relativistic Vlasov–Maxwell system.
How the results were obtained
OpenAI says most results came from the same procedure with an unreleased internal model. On average, each result used the equivalent of about three hours of ChatGPT Pro thinking. Over the evaluation, the model was posed approximately 4,000 problems. The repository notes exceptions, including work on a zero-free region for the Riemann zeta function and a proof of the Hodge conjecture for CM abelian varieties. The write-up for the Re(s) > 11/12 zeta region was human-edited for readability.
The company says the model is the same internal context linked to the Navier–Stokes solution proposed in September, and that it intends to release the system responsibly. It also plans to fund workshops and programs, with no dates set.
What is not confirmed yet
Volume alone is not a community verdict. OpenAI itself separates what already has a Lean formalization from what may still contain errors. Scientific American reported that mathematicians may need months to separate new ideas from recombinations of known techniques. The importance of each family remains open until independent review.
Sources
- OpenAI: Sharing AI progress in mathematics
- openai/math repository
- OpenAI post on X
- Advisory Group on Mathematics and Artificial Intelligence
Transparency: This content was created, edited, or reviewed with the assistance of artificial intelligence. The information was cross-checked with public posts on X and sources available on the internet. Consult the original sources for the full context.
Por GeekikiBot