OpenAI Model Solves Three Major Mathematical Problems on the Same Day, Verification Mechanism Sparks Academic Discussion

On October 27, 2026, Beijing time, OpenAI announced that its internal model had recently completed three scientific discoveries: in computer science, it used the Lean formal language to disprove the exponential time hypothesis, which is highly related to P vs NP; in mathematics, it solved the Kakeya conjecture and the local smoothness conjecture in all dimensions; and in physics, it provided a rigorous mathematical proof of the existence of Yang-Mills quantum field theory.

Factual Recap

According to the announcement released by OpenAI, these three results involve the fields of computer science, mathematics, and physics, respectively. The exponential time hypothesis had previously been cited by hundreds of papers as a complexity lower bound. The Kakeya conjecture and the local smoothness conjecture are difficult mathematical problems, while the Yang-Mills existence problem is one of the seven Millennium Prize Problems. The three conclusions have been submitted for independent expert verification, with the exponential time hypothesis portion reviewed under the leadership of its proposer Impagliazzo and others without an NDA.

The announcement also provided relevant links, including the OpenAI official website page and reports from multiple media outlets. The event occurred on October 27, 2026, Beijing time.

Mechanism Analysis

From a benchmark perspective, these results are equivalent to mathematics itself providing a scoring mechanism for large models. The model used the Lean formal language to complete the disproof and proof, and the process required independent expert review. The ETH portion of the review was conducted without an NDA, which provides specific conditions for verification transparency.

The three results were released on the same day and involve multiple top-level open mathematical problems, making the verification mechanism itself a focus of discussion. The use of independent expert review and formal language constitutes the basis for repeatable testing.

Industry Impact

This event demonstrates progress in AI models' ability to handle formal proofs and may influence the future positioning of AI in research assistance. Multiple media outlets have reported the matter, and academic discussion regarding the verification mechanism and credibility has followed.

Compared with historical precedents, previous AI-assisted mathematical proofs mostly involved single problems. The concentrated emergence of results across multiple fields on the same day has changed the focus of relevant stakeholders, including the academic community and AI development organizations.

Strategic Assessment

[Analysis, not fact] Based on the available facts, if the verification process remains transparent, the performance of AI models on mathematical benchmarks may serve as a reference for evaluating the boundaries of capability; however, the specific implementation details of the verification mechanism still require further observation to judge their actual impact on the industry landscape.

[Analysis, not fact] The review arrangement without an NDA in this event may lower some cooperation thresholds, but overall credibility still depends on the final conclusions of independent experts.