TLDROpenAI claims approximately 10,000 AI agents cracked a portion of the Navier-Stokes mathematical challenge in 88 hoursThe AI system produced 2.7 million messages and approximately 130 billion tokens throughout the workThe underlying model exceeds the capabilities of any OpenAI system currently available to the publicTristan Buckmaster, a mathematician at NYU, suggests OpenAI may have accessed details from his unpublished workOpenAI states it will not pursue the $1 million reward and the proof awaits independent validationOpenAI has announced that an experimental network of approximately 10,000 AI agents successfully tackled a portion of the Navier-Stokes existence and smoothness challenge within 88 hours. This mathematical puzzle represents one of the seven Millennium Prize Problems, with each offering a $1 million bounty from the Clay Mathematics Institute.We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics.The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra.The problem… pic.twitter.com/8zol3BPTL4— OpenAI (@OpenAI) September 8, 2026The Navier-Stokes equations model fluid dynamics for substances such as air and water. The longstanding mathematical question centers on whether smooth fluid motion can ultimately deteriorate and generate velocities that increase indefinitely. According to OpenAI’s system, such breakdown is possible.The proposed proof outlines a fluid vortex that elongates and intensifies until its velocity explodes within finite time, despite the system maintaining finite total energy throughout.Throughout the problem-solving process, the AI agents communicated via approximately 2.7 million messages and produced roughly 130 billion tokens. Using OpenAI’s standard pricing structure, such computational resources would amount to approximately $10 million in costs.The underlying system is a proprietary model that OpenAI describes as superior to GPT-6 Astra, currently its most sophisticated publicly accessible model. Following the agents’ completion of their work, Astra dedicated an additional 17 hours to verification and formalization of the proof.According to OpenAI, training for this advanced model commenced in late August. By September 1, the organization had received information suggesting that two Millennium Prize challenges had been resolved by other parties. This prompted OpenAI to direct its AI agents toward the outstanding problems.By September 5, approximately 88 hours after initiation, OpenAI had produced a candidate solution.Mathematician Questions Use of Unpublished ResearchThe revelation prompted an immediate reaction from Tristan Buckmaster, a mathematics professor at New York University. He revealed that he and Levent Alpöge, a mathematician employed at Anthropic, had been investigating a closely parallel problem for approximately one year.Buckmaster disclosed that both researchers had utilized OpenAI’s Codex platform during their investigation. On September 3, he discovered that details regarding their research progress had been transmitted to OpenAI.He asserted that OpenAI commenced work on the Navier-Stokes challenge only following receipt of that information. He further emphasized that virtually no other mathematicians were exploring the identical methodology he and Alpöge had adopted.OpenAI has rejected claims of directly accessing Buckmaster and Alpöge’s research materials. The organization stated that no user information was retrieved and that no confirmed connection exists. Nevertheless, OpenAI acknowledged it cannot entirely eliminate the possibility that data from the researchers’ utilization of OpenAI tools may have contributed to model enhancement.OpenAI extended congratulations to both researchers for their contributions, characterizing the work as simultaneous rather than preceding. The company additionally observed that the two proofs diverge in approach and actually establish somewhat distinct mathematical results.The proposed solution remains unverified by independent review. The Clay Mathematics Institute mandates that candidate solutions undergo extensive peer examination spanning multiple years before any prize distribution occurs.OpenAI has confirmed it will not pursue the $1 million reward, framing the achievement as evidence of the accelerating advancement of its research infrastructure.The post OpenAI Claims AI Breakthrough on Million-Dollar Math Challenge Sparks Controversy appeared first on Blockonomi.