Particle.news

OpenAI Says Model Solved Navier–Stokes Using Scaled Spanish Method

The claim spotlights how AI can execute human mathematical ideas and raises urgent demands for independent proof, clearer attribution, and scrutiny of model training

Overview

  • OpenAI announced that one of its models produced a claimed solution to the Navier–Stokes problem and disclosed heavy compute use, reporting 10,000 agents running for 88 hours.
  • Researchers Diego Córdoba and Luis Martínez‑Zoroa are now widely seen as the originators of the core method the model automated, and Spanish bodies have demanded explicit recognition of their contribution.
  • Tristan Buckmaster said OpenAI contacted him saying it already had a proof, offered publication options, and sought to exclude his collaborator Levent Alpöge, prompting debate about transparency and treatment of private work.
  • OpenAI said it will not claim the Clay Millennium Prize and denied access to specific user data while acknowledging it cannot rule out influence from anonymized, derived training data.
  • Mathematicians and institutions stress that the claimed proof still needs independent, rigorous validation and that the episode exposes how vast corporate compute can rapidly accelerate results and reshape credit for ideas.