OpenAI ran 10,000 agents for 88 hours to prove Navier-Stokes — and it cost millions
AGI Hunt · wechat · 2026-09-10
Per AGI Hunt's writeup, OpenAI ran 10,000 agents continuously for 88 hours to produce a proof of the Navier-Stokes equations with Lean formal verification, burning millions of dollars of compute — roughly a century of uninterrupted work by one brilliant human. When someone joked about Anthropic having found room-temperature superconductors and rivals rushing to scoop it the same way, Sam Altman replied "you know what let's try." OpenAI's blog also claims a next-gen model smarter than GPT-6 was trained in days. The author's takeaway: throwing hundreds of millions of agents at decades-old problems with machine-verifiable criteria may soon be the norm.
More from AGI Musings
- OpenAI confirms progress on second Millennium Prize problem, sparking RSI debate — AndyMasley · 2026-09-11
- OpenAI Internal Model Ran 10K Agents for 88 Hours, Claims Partial Navier-Stokes Proof — eyishazyer · 2026-09-11
- Paras Chopra: Build AI as exoskeletons for humans, not total replacements — CatAstro_Piyush · 2026-09-11
- Ex-OpenAI/Anthropic researcher resigns, says labs are gambling lives racing to self-improving superintelligence — csuwildcat · 2026-09-11
- Richard Ngo: How AI safety got captured by OpenAI and later Anthropic — RichardMCNgo · 2026-09-11
- Ex-Googler's dark take: being killed by a Chinese model feels more dignified — moultano · 2026-09-11