OpenAI Says AI Solved Navier-Stokes Millennium Problem — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI announced that an internal next-generation model "significantly more capable than GPT-6 Astra" produced a solution to the Navier-Stokes Millennium Prize Problem, one of the $1 million Clay-style challenges highlighted by multiple outlets.
- The proof was generated by 10,000 concurrent agents working for 88 hours, according to OpenAI's official X announcement, which framed the model as already exceeding the publicly known GPT-6 Astra frontier.
- The Navier-Stokes problem concerns whether the equations governing fluid flow always have smooth solutions or can develop singularities, a question unresolved in classical mathematics.
- New York Times, New Scientist, Scientific American, Axios, and Neowin all covered the announcement as a potential historic breakthrough, with Scientific American flagging it comes "amid swirl of controversy."
- Miles Brundage tweeted that it is "pretty fucked up" a model already exists "significantly more capable than GPT-6 Astra," warning "things are going much too fast."
- Nathan Calvin observed that the AI 2027 authors, who published in mid-2025, "would have been laughed out of the room" for predicting an AI would solve Navier-Stokes by September 2026.
Why it matters: The claim, if verified, would mark the first solution to a Millennium Prize Problem carrying a $1 million purse, and signals OpenAI is already operating a model substantially beyond its public GPT-6 Astra release. That internal capability gap drew explicit alarm from independent researcher Miles Brundage, whose 'things are going much too fast' quote is the most concrete stakeholder pushback in the coverage cluster.
Ask SkimNews
