OpenAI was reported on October 6, 2026, to be planning a GitHub release of hundreds of mathematical results. Separately, spokesperson Lindsay McCallum said the company’s new internal model had resolved more than 100 long-standing open problems, including the Navier–Stokes Millennium Prize problem.

OpenAI’s planned mathematics release

The reported plan called for OpenAI to publish hundreds of results on GitHub. McCallum said the company had not set a release time when she discussed the plan on October 6.

The planned release and the model’s claimed results are separate figures: the first refers to results OpenAI intended to publish; the second is the number of open problems McCallum said the model had resolved.

What OpenAI says its model resolved

McCallum said training on the new internal model began on August 28, 2026. She said it had resolved more than 100 long-standing open problems across most areas of mathematics, including the Navier–Stokes Millennium Prize problem, which concerns equations used to describe fluid motion.

Why mathematicians questioned the release approach

OpenAI convened about 40 mathematicians in August to discuss AI’s mathematical capabilities and how results should be communicated. Mathematician Bryna Kra said attendees urged the company to publish explanatory papers so other mathematicians could understand and use the work, rather than relying only on blog posts or short social media announcements. Kra said the meeting’s input appeared not to have changed OpenAI’s approach.

That debate is about more than presentation. Detailed papers give other researchers a way to examine the reasoning, build on results and recognize earlier contributions.

The Buckmaster–Alpöge authorship dispute

Tristan Buckmaster accused OpenAI of front-running related, unpublished work he had pursued with Levent Alpöge, an Anthropic employee. Buckmaster also said OpenAI researcher Sébastien Bubeck seemed to imply that Alpöge should be left off a paper. Bubeck denied asking for Alpöge to be omitted from the author list.

Tools for organizing machine-assisted mathematics

Hexagon is a repository for primarily AI-generated mathematical material. Palomar is a registry for machine-verified mathematics. The two projects serve different roles: one collects mathematical material, while the other catalogs work checked by machines.