OpenAI announced on October 6, 2026 that it is releasing a range of new mathematical results produced by what it describes as an internal frontier model. According to the company's post, it consulted the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study to develop best practices for sharing the results, and drew on that group's advice and public recommendations.
OpenAI said the results are being published in a GitHub repository with protocols for paper revisions and citations, and that it is exploring community-hosted alternatives that meet the committee's guidelines. The repository includes formalizations of many of the proofs in Lean, a programming language that allows proofs to be checked by a computer. The company said it will add more formalizations over time.
The repository also contains 10 summaries of the model's reasoning, estimates of compute spent measured in ChatGPT Pro usage, and statistics on the number of attempted problems. OpenAI stated that the average result used the equivalent compute of roughly three hours of ChatGPT Pro thinking.
OpenAI said it will fund workshops, conferences, and special programs on understanding AI-produced results, and that it is working to responsibly release the model that produced them. The announcement is a primary-source vendor post; independent verification of the results was not included.
- Results published in a GitHub repository with revision and citation protocols, dated October 6, 2026
- Proofs formalized in Lean for computer checking, plus 10 reasoning summaries
- Average result used roughly three hours of ChatGPT Pro thinking equivalent
- OpenAI consulted the IAS Advisory Group on Mathematics and Artificial Intelligence
What it means for you
OpenAI says one of its unreleased models produced new math results, and it has published the proofs along with details on how it got them. For most people and businesses this is a research milestone, not a product you can use — the model itself isn't available. The main reason to notice is that the results include machine-checkable proofs (via Lean), which is a more honest way to show AI math claims than a press release alone.
Who should care
Research mathematicians, people working in formal verification or proof-checking, and anyone tracking how far frontier models have moved beyond chat-style tasks into rigorous science.
Skip this if
You run a business or use AI for everyday work — there's nothing to adopt here, and the model that produced these results isn't released yet.
Sources: OpenAI — read the original