# Unforced Navier–Stokes blowup — Buckmaster and Alpöge post an AI-aided proof

> Two mathematicians claim the force-free Navier–Stokes blowup that OpenAI's September proof left open, written with AI help.

- Org: Tristan Buckmaster & Levent Alpöge
- Date: 2026-11-20
- Category: paper, showcase
- Source: https://github.com/tristanbuckmaster/navier-stokes
- Canonical page: https://ai-tldr.dev/releases/buckmaster-alpoge-unforced-navier-stokes-blowup/

## In short

Tristan Buckmaster (NYU) and Levent Alpöge (Anthropic) posted a 79-page, heavily AI-assisted proof that 3D Navier–Stokes flow can blow up with no outside force, with a Lean 4 formalization. The key numerical certificate is not public yet.

## What it is

Finite-time blow-up for the Navier–Stokes equations is a 79-page paper by Levent Alpöge and Tristan Buckmaster, dated November 20, 2026. It claims that 3D incompressible Navier–Stokes flow can blow up in finite time with no external force. OpenAI's September 2026 proof needed a finely tuned outside force, which many experts saw as a loophole. A companion paper with Matei Coiculescu claims the same for the frictionless 3D Euler equations.

## How it works

Terence Tao summarizes the strategy: numerically build an approximate, discretely self-similar solution, then use rigorous perturbation theory to turn it into an exact one. The existence step is computer-assisted, with an interval-arithmetic check that uses the Arb library. The rest of the argument is formalized in Lean 4 on top of Mathlib, and the repository holds the paper and the Lean package.

## Why it matters

Force-free blowup is the version of the Millennium Prize question that experts said OpenAI did not settle, so a correct proof would close the debate from September. Buckmaster calls the write-up unpolished and heavily LLM-assisted, and Tao says more work is needed for a professionally written proof. The data certificate and interval computation are not in the repository yet, so nobody outside the team can check the full result.

## Who it's for

mathematicians, formal-verification and AI-for-math researchers

## Try it

Read the paper and Lean package at https://github.com/tristanbuckmaster/navier-stokes

## Quick facts

- Claimed result: Finite-time blowup, unforced 3D Navier–Stokes
- Authors: Levent Alpöge (Anthropic), Tristan Buckmaster (NYU)
- Paper: 79 pages, dated Nov 20, 2026
- Formal proof: Lean 4 + Mathlib, on GitHub
- Companion result: Unforced 3D Euler blowup, with Matei Coiculescu
- Status: Unreviewed; certificate not yet released

## FAQ

### How is the Buckmaster–Alpöge proof different from OpenAI's Navier–Stokes proof?

OpenAI's September 2026 Navier–Stokes result showed blowup only when a finely tuned outside force drives the fluid, which Scientific American calls a loophole in the problem's wording. The Buckmaster–Alpöge paper claims blowup with no forcing at all, the version experts considered the real question. Early reports suggest OpenAI's proof took a different route, and the two sides are still in a public dispute over credit.

### Has the unforced Navier–Stokes blowup proof been verified?

Not yet. As of November 20, 2026, no independent expert had confirmed the Buckmaster–Alpöge Navier–Stokes claim. The repository includes the paper and a Lean 4 formalization, but not the data certificate or the interval-arithmetic computation that the existence step depends on. Scientific American reports the authors say the certificate will be released soon.

### What role did AI play in the Navier–Stokes blowup paper?

Terence Tao describes the Navier–Stokes and Euler papers as written with heavy AI usage, and Buckmaster calls the Navier–Stokes write-up unpolished and heavily LLM-assisted. The authors say they used Anthropic's Claude and OpenAI's Codex throughout. The Lean formalization is what lets others check the argument by machine instead of trusting the model-written prose.

### Why did Buckmaster and Alpöge release the Navier–Stokes paper before it was polished?

Tristan Buckmaster told Scientific American that after last time he and Levent Alpöge could not risk waiting. In September 2026 OpenAI announced its forced Navier–Stokes result a day after the pair posted related work, and Buckmaster has said OpenAI scooped them, which OpenAI denies. Tao says more work is needed to produce a proper, professionally written proof.

## Links

- [Paper + Lean Repo](https://github.com/tristanbuckmaster/navier-stokes)
- [Unforced Euler Repo](https://github.com/mcoiculescu/EulerBlowupPublic)
- [Terence Tao's Notes](https://terrytao.wordpress.com/2026/11/20/unforced-euler-singularity-unforced-navier-stokes-singularity-and-the-generalized-riemann-hypothesis/)
- [Scientific American](https://www.scientificamerican.com/article/ai-just-solved-the-real-navier-stokes-problem/)
- [OfficeChai Report](https://officechai.com/ai/tristan-buckmaster-levent-alpoge-claim-proof-of-navier-stokes-millennium-prize-problem-without-forcing/)
- [HN Discussion](https://news.ycombinator.com/item?id=50578358)
