OpenAI reports that its in-house model cracked more than 100 longstanding math challenges after only one month of training

OpenAI has announced a new internal model that has reportedly solved more than 100 long-standing mathematics problems in roughly a month of training, including the Navier-Stokes Millennium Problem and, according to the company, a second Millennium Prize Problem, the Hodge conjecture. The announcement came alongside the creation of an independent advisory group of leading mathematicians — a move that reads as a direct response to mounting criticism from the mathematical community about how AI progress in the field is being measured and shared.

A month of training, a century of problems

According to OpenAI, training on the model only began on August 28, meaning the entire run lasted about a month. The company says even its own mathematicians were "surprised" by the speed of progress, and that internal discussions have turned to how to give the academic community enough lead time to prepare.

The announcement marks a notable shift in direction. Chief Scientist Jakub Pachocki recently said the team had deliberately chosen not to optimize for mathematics, focusing instead on recursive self-improvement. OpenAI has also admitted it only took on the Millennium Prize Problem after hearing rumors that another team — partly made up of Anthropic researchers — had already cracked it. Skeptics may see the publicity around famous solved problems as a way to reassure current investors and draw in new ones.

Which problems the model actually solved, how it solved them, and what the results mean for math and science all remain open questions. The published Navier-Stokes solution has already triggered heated debate in parts of the scientific community.

Responding to the mathematicians' letter

The advisory group is OpenAI's answer to criticism from mathematicians, including an open letter titled "A Severe Misalignment of AI in Mathematics." In it, the signatories warn against judging AI performance by the number of open problems a system can solve. Mass-producing solutions, they argue, undermines the conceptual understanding that is the true point of mathematics — and ultimately threatens human intellect.

OpenAI says it is now working with mathematicians who have formed the Advisory Group on Mathematics and Artificial Intelligence, based at the Institute for Advanced Study. The group is intended to serve as a bridge between the company, the math community, and the wider public. OpenAI frames the collaboration as a "first step," noting there are plenty of "difficult questions ahead about how AI can support mathematical understanding and how the benefits of these capabilities can reach the wider community." The company argues such guidance is needed because mathematical breakthroughs can carry far-reaching practical consequences, making responsible development a concern well beyond the field itself.

Independence — within limits

OpenAI says the group will operate independently. Members can offer unprompted advice, speak publicly about the company's influence on mathematics, and publish their recommendations. They are not paid by OpenAI, and they control who joins.

But there is a clear boundary: the group has no say over the pace of OpenAI's research. "Importantly, the group will not be responsible for advising us on how to pace our internal progress on mathematics," the company writes. Members can influence how results are communicated — not whether or how quickly they are produced.

Gowers joins, but won't sign the letter

Among the group's founding members is Timothy Gowers, the Fields Medalist who declined to sign the open letter from 25 Fields Medalists and explained his reasoning on his blog. Gowers agrees with much of the letter and believes mathematics is in genuine trouble. His disagreement lies with how the letter frames the discipline's purpose. The letter treats conceptual understanding as the primary goal and problem-solving as merely instrumental. Gowers instead sees a spectrum: "At one end of the spectrum you have mathematicians who are primarily motivated by the wish to solve problems, who see conceptual understanding as a very important means to that end. At the other you have mathematicians who are primarily motivated by the wish to attain conceptual understanding, who see problem-solving as a very important means to that end."

Gowers also notes that his own automated theorem proving project at Cambridge lost its reason for existing once large language models became capable enough. "To put it another way, we have had to swallow the bitter lesson (which of course we were always aware was a distinct possibility, even if the speed at which it happened has taken us by surprise)."

The deeper worry: losing the next generation

What concerns Gowers most is that the social structures holding mathematical knowledge together could fall apart. People who might once have pursued a Ph.D. and become "custodians of the mathematical tradition" may conclude the path is no longer worth taking. "Speaking for myself, my main motivation for becoming a mathematician was the dream that I would solve unsolved problems — the more famous the better." Remove that dream, and nothing obvious replaces it.

Funding adds another layer of risk: policymakers could conclude that human mathematicians are redundant once AI handles the proofs. "We urgently need to come up with good ways of explaining the value of having a large pool of human mathematical experts, even if it is no longer part of their role to find new proofs of theorems," Gowers writes.

He also stayed off the letter because he couldn't identify what it was demanding that wasn't already happening. He expects publicly available LLMs capable of tackling major math problems within months. "So I felt that there was nothing to be gained from criticizing AI companies for generating too many solutions too quickly."

What to watch next is whether the advisory group gains any real influence over how — and how fast — OpenAI publishes results, and whether the full details of the claimed solutions withstand scrutiny from the wider mathematical community.

Meta description: OpenAI says a new internal model solved over 100 open math problems in a month, and formed a mathematician advisory group amid growing criticism of its approach.

Tags: OpenAI, mathematics, AI research, Timothy Gowers, Millennium Prize Problems

Featured image: Abstract technology style — flowing geometric lines and glowing nodes suggesting mathematical structure, no real people, no logos, no text.