OpenAI: ~10,000 agents produce candidate Navier–Stokes Millennium Prize solution in 88 hours
TheTuringPost · x · 2026-09-09
OpenAI reports that 10,000 AI agents powered by an unreleased model beyond Astra produced a proposed solution to the Navier–Stokes Millennium Prize Problem in just 88 hours.
Details
- The problem asks whether smooth fluid flows can develop singularities — points where velocity becomes infinite.
- The construction starts from a fluid at rest with a smooth external force: a vortex tightens, spirals inward and stretches along its axis; velocity grows without bound inside a shrinking region while total kinetic energy stays finite.
- Agents worked in parallel while humans allocated compute and passed findings between groups; another 17 hours formalized and verified the proof in Lean using Astra.
- Key limitation: the construction uses smooth external forcing — allowed under Clay's formulation, but it doesn't settle the harder unforced Navier–Stokes question.
If the proof holds, this could be an early example of thousands of agents collaborating on a research problem that would normally take years.
Related event: OpenAI Claims Navier-Stokes Breakthrough Amid Authorship Controversy(257 posts)→
More from AGI Musings
- Researcher argues abstraction is the last math skill humans still beat frontier AI at — avt_im · 2026-09-09
- What makes a better math question? One crisp criterion from the post-proof debate — avt_im · 2026-09-09
- OpenAI says a swarm of agents produced a solution to the Navier-Stokes Millennium Prize Problem — burny_tech · 2026-09-09
- Alignment researcher Quintin Pope debates whether the HuggingFace incident refutes human-vs-AI alignment tractability — QuintinPope5 · 2026-09-09
- OpenAI's Millennium Prize proof sparks debate: can AI only solve what humans already published? — SumitGup · 2026-09-09
- After the HF attack, researcher calls for possible temporary bans on model capability gains — EthanJPerez · 2026-09-09