ChatGPT News 09/08/2026 AI Rating: High

OpenAI Shares an AI-Generated Partial Solution to the Navier-Stokes Millennium Prize Problem

#OpenAI#Mathematics#NavierStokes#MillenniumPrize#AIforScience

OpenAI shared an AI-generated analytical proof addressing part of the Navier-Stokes Millennium Prize Problem, produced by an internal, unreleased model more capable than GPT-6 Astra that began training on August 28, 2026.

Details

  • What was proven: the proof shows that a smooth three-dimensional incompressible fluid, initially at rest, can develop a singularity — infinite velocity — in finite time, despite a smooth applied force and finite energy throughout, resolving statements “C” and “D” of the Clay Mathematics Institute’s official problem formulation by disproving the smoothness conjecture
  • The technical picture: OpenAI describes the singularity as “a vortex, a spinning swirl of fluid, that spirals inward and gets increasingly elongated, like spaghetti,” requiring competing terms in the Navier-Stokes equations to both grow large and cancel out in a precise way
  • Formal verification: after the underlying result was reached on September 5, GPT-6 Astra spent an additional 17 hours completing a formal verification of the proof in the Lean theorem-proving language
  • Scale of the effort: OpenAI says the resolution involved roughly 10,000 concurrent agents, 2.7 million messages, and 130 billion output tokens
  • Explicit caveat: OpenAI states directly, “we do not intend to claim the Millennium Prize for this result,” and notes concurrent independent work on related problems by Anthropic researcher Levent Alpöge and NYU’s Tristan Buckmaster
  • No independent peer review yet: the announcement does not mention review or confirmation by the outside mathematical community

What happened next

The result showcases capabilities from a model beyond GPT-6 Astra that OpenAI hasn’t released, using a scale of parallel agentic compute — thousands of agents, millions of messages — well beyond typical single-session model use. Given the explicit disclaimer about the Millennium Prize and the lack of outside peer review so far, the mathematical community’s independent verification of the proof will determine how the result is ultimately received.