Fact-checking the Lean repo: OpenAI proved Navier–Stokes alternative (C), not the whole prize
CatAstro_Piyush · x · 2026-09-09
A technical blog dissects the claim "OpenAI solved Navier–Stokes." The author's first reaction was to ask what exactly the prompt was; his second was to read the five-page problem statement Charles Fefferman wrote for the Clay Mathematics Institute in 2000 before touching any news coverage. His verdict on the three jobs that sentence does: (1) a theorem about 3D Navier–Stokes with smooth external force was proved and checked by a proof assistant — as far as anyone can tell, correct (the author cloned the Lean repo to verify); (2) it matches, clause for clause, alternative (C) of the official statement, and by a small extension (D) — verifiable yourself since the clause is two sentences; (3) "that alternative is what people mean by the Navier–Stokes problem" — this is false, in a way Fefferman anticipated 26 years ago when he wrote four alternatives instead of one. The post is an argument for reading the official definition before the headlines.
Related event: AMS Confirms Navier-Stokes Breakthrough as Lean Proof Goes Open Source(4 posts)→
More from Research
- DONIC hits 86.79% on bimanual dexterous tracking, crushing prior SOTA's 9.43% — liuziwei7 · 2026-10-27
- Why don't more LLMs ship 10M-token context? A researcher asks about attention tradeoffs — agarwl_ · 2026-10-27
- PSNR below 15 dB in papers likely means the metric was misused, researcher warns — kwangmoo_yi · 2026-10-27
- Ortet's first health eval: 34 open-weight models, 50 tasks, generic rankings fall short — thegautamkamath · 2026-10-27
- Scaling-fit discrepancy: output-head FLOPs are ~65% of compute at the smallest rung, ~3% at 535B — dlwh · 2026-10-27
- Marin's dlwh Explains the Scaling Law Fit: Estimating the Loss Floor Early — dlwh · 2026-10-27