Tristan Buckmaster and Levent Alpöge announced a claimed finite-time blowup result for the unforced three-dimensional incompressible Navier–Stokes equations on November 20, 2026. Their release includes a 79-page paper and a Lean 4 formalization. The distinction from OpenAI’s September result is important, but the materials available here do not establish that the new proof has been independently validated.
3
6
13
What the new claim changes
OpenAI’s September result concerned Navier–Stokes equations with a smooth external force; coverage described it as involving a fluid initially at rest and addressing statements C and D of the Millennium Prize formulation.
4
12 Buckmaster and Alpöge’s November claim instead concerns the unforced equations, with smooth initial data.
6
13
That difference matters mathematically: the new claim is that singularity can form without an imposed force driving the fluid. It is more precise to describe this as a stronger, different result than to say OpenAI’s announcement violated the prize wording: the cited account says OpenAI’s result addressed the forced formulation.
4
What has been released—and what remains open
The authors’ repository lists Finite-time blow-up for the Navier–Stokes equations, a 79-page version dated November 20, and its accompanying Lean 4 formalization.
3 A report on the release says the numerical certificate associated with the result was not yet available.
13
A paper and formalization are significant research artifacts, but their publication alone does not establish independent mathematical acceptance. The sources available here do not document a completed independent review or recognition by the Clay Mathematics Institute. The missing certificate also limits what can be checked from the release described in the reporting.
3
13
What can—and can’t—be said about the proof strategy
The supplied sources establish the headline claim, paper, formalization, and reported certificate gap, but they do not provide enough detail to explain the numerical construction or its rigorous perturbation argument reliably. A numerical approximation by itself would not establish blowup; the decisive question is whether the paper proves that an exact solution with the claimed behavior follows from that construction. The evidence here is not enough to assess those technical steps.
Terence Tao’s November announcement identifies the Navier–Stokes result as unforced and heavily AI-assisted. It also describes a separate unforced Euler result as human-readable. The available excerpts do not provide Tao’s detailed assessment of the Navier–Stokes proof’s exposition or strategy, so those views should not be inferred.
8
The takeaway
Buckmaster and Alpöge’s announcement addresses a meaningful distinction: their claimed result removes the external force present in OpenAI’s September result. The 79-page paper and Lean 4 code make the work available for scrutiny, but the evidence cited here does not settle whether the proof is correct or independently accepted.
3
4
13