OpenAI Cracks a Century-Old Math Puzzle, Igniting a Fierce Priority Dispute

Key points

OpenAI deployed an internal model and roughly 10,000 autonomous agents to produce a proposed proof for the 90-year-old Navier-Stokes existence and smoothness problem ¹ ² ⁴.
The 100-page proof asserts that a fluid vortex can develop a singularity, meaning its speed grows infinitely within a finite time while keeping its total energy finite ¹⁵ ²⁰.
NYU mathematician Tristan Buckmaster publicly challenged the claim, alleging OpenAI rushed to finalize its proof after learning of his and Anthropic researcher Levent Alpöge's unpublished work ² ⁶ ²⁰.
OpenAI denied using any specific user data or unreleased rival research, while stating it has no intention of claiming the $1 million Millennium Prize from the Clay Mathematics Institute ² ¹⁰ ¹⁷.

Swarm Intelligence Meets Fluid Dynamics

In a milestone that underscores the rapid encroachment of artificial intelligence into pure mathematics, OpenAI announced that an unreleased internal system solved the Navier-Stokes existence and smoothness problem ¹ ¹⁵. The puzzle, which has stumped human mathematicians for roughly nine decades, is one of the seven prestigious Millennium Prize Problems established by the Clay Mathematics Institute in the year 2000, each carrying a $1 million bounty ² ⁴ ¹⁰. Rather than relying on a single model reasoning at length, the breakthrough was achieved by deploying a coordinating swarm of approximately 10,000 autonomous AI agents working in parallel ² ⁴ ¹⁷. Operating over an intense 88-hour span, the agent collective exchanged millions of messages and consumed billions of tokens before generating a detailed, 100-page proposed proof ⁴ ¹⁵ ²⁰. OpenAI’s public GPT-6 Astra model was subsequently put to work for another 17 hours to help formalize and check the resulting arguments ² ¹⁵.

The core mathematical finding focuses on whether smooth three-dimensional fluid motion can break down under physical laws ¹. The Navier-Stokes equations describe how fluids like air and water move through a continuous medium, governing everything from aircraft design to global weather forecasting ¹ ⁴. OpenAI’s proof demonstrates that under specific conditions involving a smooth external force, an initially smooth fluid at rest can develop a singularity in finite time ¹ ¹⁷. In practical terms, the equations suggest that a fluid vortex can become so stretched and concentrated that its speed blows up to infinity while its total energy remains finite ¹⁵ ²⁰. Because true physical fluids cannot move at infinite speeds, the result implies that the continuum approximation of the Navier-Stokes equations breaks down at extreme scales ¹ ¹⁵.

Verification and the Choice to Forego the Prize

To ensure the mathematical community could verify the validity of the work without simply taking the company’s word for it, OpenAI published a full write-up, a detailed PDF paper, and a complete formalization of the argument in Lean ¹ ¹⁷. Lean is a specialized programming language and computer proof assistant that allows every logical step of a mathematical proof to be rigorously checked by machine, stripping away subjectivity ¹⁷ ²⁰. This formal verification step is becoming increasingly vital as AI models churn out complex mathematical reasoning faster than human experts can audit by hand ²¹.

Despite the monumental nature of the achievement, OpenAI made the striking announcement that it does not intend to claim the $1 million Millennium Prize associated with the problem ² ¹⁷. Under the rules of the Clay Mathematics Institute, proposed solutions must undergo years of rigorous peer review and broad scrutiny before gaining official recognition ¹⁰ ¹⁵. By choosing not to formally submit for the bounty, OpenAI signaled a pragmatic awareness of the high bar required for acceptance, leaving the ultimate validation of its forced-flow methodology in the hands of the global mathematical community ¹⁷.

Academic Clash and Questions of Provenance

The celebratory atmosphere surrounding the announcement was quickly overshadowed by controversy over how the proof came to be ² ²⁰. Shortly before OpenAI went public, New York University mathematics professor Tristan Buckmaster published a detailed statement alleging that the company had accelerated its research after learning about his own unpublished work ² ⁶. Buckmaster revealed that he and Levent Alpöge, a mathematician who is employed by rival lab Anthropic, had spent nearly a year pursuing a specialized and uncommon route toward the fluid-dynamics problem, utilizing AI tools including OpenAI’s Codex during their collaboration ⁶ ²⁰. Buckmaster questioned whether data or insights from their coding sessions could have inadvertently leaked or influenced OpenAI’s internal models, raising sensitive questions about data governance and the training lineage of frontier systems ⁶ ¹⁵.

OpenAI strongly pushed back against the allegations of intellectual theft ¹⁰ ²⁰. During a press briefing, company researchers maintained that their internal system arrived at the solution independently, noting that their approach relied on a forced-fluid variant that differed from the unforced Euler equations tackled by the two mathematicians ⁶ ¹⁰. While OpenAI acknowledged that it could not completely rule out the possibility that de-identified telemetry data from general product usage might have indirectly improved its models over time, it insisted that no specific user files or private research prompts were ever accessed before publication ⁴ ⁶ ¹⁵.

A Paradigm Shift for Mathematical Research

Separately, the broader landscape of automated reasoning experienced another seismic jolt when Anthropic announced that its Claude AI had successfully produced the first fully computer-checked proof of Fermat’s Last Theorem, generating 13 million lines of Lean verification code in just 11 days ²¹. Together, these rapid developments highlight a profound turning point for academic research ¹⁸ ²¹. The transition from slow, human-led verification projects spanning years to automated agent swarms capable of tackling century-old problems in days is reshaping how mathematics is practiced ¹⁷ ²¹.

Yet this technological leap brings complex new challenges to the scientific community ¹⁴ ¹⁷. As prominent mathematician Terence Tao and other observers have noted, the rise of mass-scale, AI-generated proofs risks overwhelming human researchers with massive, opaque blocks of machine code that are extraordinarily difficult to audit for intuitive understanding ¹⁴. Furthermore, when breakthrough discoveries require millions of dollars in compute and fleets of thousands of autonomous agents, mathematical research risks becoming concentrated entirely within a handful of heavily capitalized technology corporations ¹⁷.

Companies mentioned: OpenAI • Anthropic

Primary sources

On the Navier–Stokes Millennium Prize Problem | OpenAI (openai.com) – OpenAI's official research announcement shares a formal write-up and Lean proof for the Navier-Stokes existence and smoothness problem, demonstrating that fluid motion can develop a singularity in finite time using an internal model more capable than GPT-6 Astra. Published by OpenAI on September 8, 2026.
OpenAI claims to have solved maths problem that stumped humans for decades (theguardian.com) – The Guardian reports on OpenAI's claim that 10,000 autonomous AI agents solved the Navier-Stokes problem in 88 hours, detailing mathematician Tristan Buckmaster's concerns regarding potential exposure of his unpublished work through Codex, alongside OpenAI's denial and decision not to claim the $1 million prize. Written by outside reporting in The Guardian.

Further sources

OpenAI says 10,000 AI agents cracked one of math's hardest problems in 88 hours (phys.org) – Phys.org covers OpenAI's announcement that an unreleased AI model and 10,000 agents solved a major mathematical puzzle, sparking a priority dispute with human researchers. Published by Phys.org.
OpenAI claims to have solved the 90-year-old Navier-Stokes math problem in 88 hours (cnbc.com) – CNBC details OpenAI's release stating that 10,000 coordinating agents equipped with internet and code-execution tools solved the Navier-Stokes problem in 88 hours, while highlighting pushback from NYU mathematician Tristan Buckmaster regarding his collaborative work with Levent Alpöge. Published by CNBC on September 8, 2026.
OpenAI claims solution to one of math’s $1 million Millennium Prize problems (washingtonpost.com) – The Washington Post reports on OpenAI's release of a technical paper addressing one of the $1 million Millennium Prize Problems, noting it will face intense scrutiny from global mathematicians. Published by The Washington Post.
OpenAI fought dirty on career-making math problem (techcrunch.com) – TechCrunch explores the academic controversy surrounding NYU professor Tristan Buckmaster and Anthropic mathematician Levent Alpöge's parallel findings on fluid dynamics, detailing allegations that OpenAI leveraged insights from their Codex usage to race toward a formal proof. Published by TechCrunch.
What's going on with OpenAI and the Navier-Stokes controversy? (engadget.com) – Engadget briefly notes the murky credit surrounding AI's purported solution to a major mathematical problem. Published by Engadget.
Why is there controversy around OpenAI’s Millennium Prize maths breakthrough? (newscientist.com) – New Scientist examines the broader context of AI accelerating fluid dynamics research via the Navier-Stokes puzzle amidst emerging controversies. Published by New Scientist.
How An AI math breakthrough ignited a controversy (science.org) – Science provides brief commentary on how the recent AI math breakthrough ignited community debate. Published by Science.
OpenAI claims blockbuster math breakthrough amid swirl of controversy (scientificamerican.com) – Scientific American reports on OpenAI's proof showing that Navier-Stokes equations can blow up into infinite speeds, detailing initial reactions from original forcing theorists Diego Córdoba and Luis Martínez-Zoroa alongside Sébastien Bubeck's denials of prompt contamination. Published by Scientific American on September 8, 2026.
OpenAI has solved the Navier-Stokes Millennium problem using $15m of AI effort (newscientist.com) – New Scientist highlights OpenAI's claim of solving the Navier-Stokes Millennium problem using massive computational effort, noting the surrounding priority disputes. Published by New Scientist.
Major breakthrough made on famous Millennium maths problem (newscientist.com) – New Scientist reports on collaborative human-AI findings bringing researchers closer to solving the Navier-Stokes puzzle. Published by New Scientist.
Terence Tao (@tao@mathstodon.xyz) (mathstodon.xyz) – Terence Tao shares mathematical commentary on Mastodon detailing Alpöge and Buckmaster's approach to constructing blowup solutions for fluid equations using high-frequency corrections and Lean formalization. Published on Mastodon by Terence Tao.
Terence Tao (@tao@mathstodon.xyz) (mathstodon.xyz) – Terence Tao discusses on Mastodon how primarily AI-generated solutions to global regularity problems could potentially inhibit future theoretical advances and field development despite computational successes. Published on Mastodon by Terence Tao.
OpenAI says 10,000 AI agents solved a $1 million math problem. Now mathematicians are fighting (coindesk.com) – CoinDesk details how roughly 10,000 AI agents exchanged 2.7 million messages to solve the Navier-Stokes problem in 88 hours, examining the physical implications of fluid vortices reaching infinite speed and detailing Tristan Buckmaster's data-leak concerns. Published by CoinDesk.
Terence Tao, UCLA (math.ucla.edu) – Terence Tao's professional academic directory page at UCLA providing general contact and publication links. Published by UCLA.
OpenAI publishes its Navier-Stokes proof and says it will not claim the Millennium Prize (thenextweb.com) – The Next Web analyzes OpenAI's published paper and Lean formalization repository, noting that the proof relies on an applied smooth force and explaining why OpenAI's decision not to claim the $1 million prize signals uncertainty regarding the Clay Institute's strict criteria. Published by The Next Web.
OpenAI agents find proof to $1 million Millennium Prize Problem (semafor.com) – Semafor outlines OpenAI's deployment of 10,000 agents to solve the Navier-Stokes problem, framing it as a major historical milestone while noting the accompanying researcher drama. Published by Semafor.
OpenAI Says Its AI Agents Solved Prize Math Problem (theinformation.com) – The Information reports on OpenAI's announcement that its AI agents successfully solved a prominent Millennium Prize math problem. Published by The Information.
OpenAI Says It Solved a $1M Math Problem. A Rival Mathematician Says He Did It First (decrypt.co) – Decrypt covers OpenAI's Lean-verified Navier-Stokes proof alongside detailed accounts of Tristanヴォ Buckmaster's accusations against Sébastien Bubeck regarding unpublished research, including text exchanges and executive responses from Sam Altman. Published by Decrypt.
$math 1500720 1920 gID 4 OpenAI Cracks a Century-Old Math Puzzle, Igniting a Fierce Priority Dispute$ AI Just Solved a 350-Year-Old Math Problem By Writing the Longest Proof Ever (decrypt.co) – In a related development, Decrypt reports that Anthropic's Claude AI produced the first fully computer-checked proof of Fermat's Last Theorem in 11 days, writing 13 million lines of Lean code with the help of specialized coordination tools like Prove2Me. Published by Decrypt.