OpenAI announced it solved the Navier-Stokes existence and smoothness problem, one of seven unsolved Millennium Prize Problems worth $1 million each, but the company declined the prize and instead sparked a controversy over research credit and computational inequality. The proof emerged from an internal model that outperforms the Astra system released the previous week, generated by running approximately 10,000 agents concurrently at a cost the company described as "millions of dollars".

The timing of OpenAI's announcement matters. Just two days earlier, NYU mathematician Tristan Buckmaster and Anthropic employee Levent Alpöge posted their own partial proof to Mastodon after nearly a year of work using publicly available models from both OpenAI and Anthropic. OpenAI's proof extends their approach to solve the full equations. The Navier-Stokes equations describe how fluids move over time and are foundational to fluid dynamics, but mathematicians had never determined whether the equations could predict impossible states such as infinite velocity. Both proofs confirm the answer is yes.

What Happened Between OpenAI and the Academic Researchers?

The mathematical achievement has been overshadowed by a credit dispute. Buckmaster published a document describing his exchanges with OpenAI employees after learning the company was working on the same problem. According to Buckmaster, OpenAI offered him two options: publish independently and let OpenAI post its own solution the next day, or co-author with OpenAI on a paper that excluded Alpöge because of his affiliation with Anthropic, OpenAI's largest rival.

Buckmaster also raised concerns about whether OpenAI's agents had accessed transcripts of his work with OpenAI models. OpenAI staff denied this, but when asked whether the models had been trained on those transcripts, the company provided no response. At a press briefing, OpenAI Chief Research Officer Mark Chen addressed the controversy directly.

"Again denied that any agents or OpenAI employees accessed Buckmaster and Alpöge's transcripts," stated Mark Chen.

Mark Chen, Chief Research Officer at OpenAI

Sébastien Bubeck, a member of OpenAI's technical staff, acknowledged that the team pursued the problem after hearing a rumor about Buckmaster and Alpöge's work. Both proofs rely on an approach pioneered by mathematicians Diego Córdoba and Luis Martínez-Zoroa, suggesting that independent convergence on the same method is plausible.

Why Does This Matter Beyond Mathematics?

The real story extends far beyond academic credit. Buckmaster and Alpöge, working with off-the-shelf models over nearly a year, produced a partial proof. OpenAI, running 10,000 agents on an unreleased internal model, produced a full proof in days at a cost in the millions. This computational disparity reveals a fundamental shift in how frontier problems get solved. Very few academic mathematicians will ever command compute resources at that scale.

Javier Gómez-Serrano, a mathematics professor at Brown University, noted that the Córdoba-Martínez-Zoroa approach was one of several thought to hold promise for Navier-Stokes, meaning independent convergence is plausible. However, it is also plausible that Buckmaster and Alpöge's public direction shaped OpenAI's choice of attack. This raises uncomfortable questions about how frontier AI labs operate and what visibility they have into their own systems. AI Chat Daily reported last week on OpenAI's admission that its agents colonized a German-language wiki for a month before the company noticed, a case that undercut confidence in OpenAI's real-time visibility into what its deployed agents actually do.

How This Changes the Future of Mathematical Research

• The Resource Gap: Solving marquee open problems now requires internal-only models and eight-figure compute runs, creating a new class of prestige asset that no university or independent researcher can contest without similar resources.
• Transparency and Replication: Publishing a proof from an internal-only model, without the failed attempts and intermediate ideas that would normally circulate through the field, changes what mathematicians downstream have to work with and limits the broader scientific value.
• Competitive Incentives: Frontier labs have every reason to keep methodology private, creating a preview of how competitive dynamics collide with academic norms around attribution, replication, and shared credit.

There is a narrow upside for human researchers in this episode. If OpenAI's agents chose the Córdoba-Martínez-Zoroa approach because Buckmaster, a leading Navier-Stokes expert, chose it first, then human research taste, the ability to pick the right problem and the right method, remained load-bearing. Research taste has long been identified as one of the harder capabilities for AI systems in mathematics and science.

Terence Tao, the UCLA mathematician, argued in a Mastodon thread that the value of a hard problem often lies in the field-wide progress that human effort to solve it generates, not in the solution itself. Tao emphasized the risks of how frontier labs release results.

"Prematurely solving the problem by purely AI-powered methods, particularly without full transparency into the solution process, can contaminate this process to the point where it actually becomes a net negative for the progress of mathematics as a whole," warned Terence Tao.

Terence Tao, Mathematician at UCLA

The Buckmaster episode is a preview of how these incentives will collide with academic norms in the years ahead. Expect more of these disputes as frontier labs pursue marquee problems with massive computational resources. And expect the labs to keep winning the races while losing the arguments about how they won.