OpenAI publishes 372 AI-generated mathematical proofs on GitHub

OpenAI released 372 mathematical results generated by an internal frontier model, each claimed to solve an open problem or make substantial progress, including improvements to an important computer algorithm and progress related to the Riemann hypothesis. The results are hosted in a GitHub repository with a revision log and citations, and some proofs include Lean formalization to enable machine verification and ease the burden of human review. OpenAI says nearly every result came from a single prompt to a single AI agent, consuming on average about three hours of ChatGPT Pro Thinking compute, whereas a previous solution to a Navier-Stokes problem required a cluster of 10,000 agents and millions of dollars in compute. The company did not disclose the prompts, only the average compute cost, and said it consulted the Institute for Advanced Study's Mathematics and AI Advisory Group, but stipulated in advance that the group could only advise on how results should be disseminated, not on whether or how quickly results should be produced.