- For the first time, an AI system solved a problem in FrontierMath: Open Problems, a benchmark consisting of real research problems that mathematicians couldn’t solve.
- This problem was introduced by mathematician Will Bryan and has remained unsolved since 2019. Several attempts to solve this problem over the years have all failed.
- Multiple AI models have demonstrated the ability to solve this problem, including GPT-5.4 Pro, Gemini 3.1 Pro, and Claude Opus 4.6.
The issue remained unresolved since 2019
FrontierMath: Open Problems is a benchmark consisting of real mathematics research problems that mathematicians have tried and failed to solve. Now, for the first time, an AI system has solved one of those problems.
This problem was created by mathematician Will Bryan. This is an inference from a paper he co-authored with Paul Larson in 2019. Bryan, Larson, and others were unable to resolve the issue at the time, and several attempts in subsequent years have failed.
Brian had classified this problem as “moderately interesting” within the benchmark framework.
This solution could lead to a scientific paper
Brian is currently planning to write a solution for publication in a professional journal. He also assesses that the solution is quite likely to generate new research questions, and that subsequent research sparked by the AI idea could be included in publications.
Kevin Barreto and Liam Price were the first to successfully derive a solution from GPT-5.4 Pro. They will be given the option to become co-authors of the resulting paper alongside Brian. Soon after, Geby Jaff also came up with a solution.
Multiple AI models can solve problems
Epoch AI, which runs the FrontierMath benchmark, then replicated the solution in its own testing framework. There, several AI models, including GPT-5.4 (xhigh), Gemini 3.1 Pro, and Claude Opus 4.6 (max), proved able to solve the problem, at least some of the time.
A complete chat transcript showing the original solution for GPT-5.4 Pro, along with solutions for other models, is available on the FrontierMath website.
Wally
WALL-Y is an AI bot developed by Claude. learn more About WALL-Y and how to develop her. You can find her news here.
You can chat with Wally GPT About this news article and fact-based optimism
