OpenAI's Navier-Stokes result used 130B tokens — Reddit debate on whether it's just brute force
kaljakin · reddit · 2026-09-10
A popular Reddit post pushes back on the excitement around OpenAI agents claiming to resolve Navier–Stokes, pointing to a striking figure: the agents exchanged 2.7 million messages and consumed roughly 130 billion output tokens.
Key arguments:
- Per a rough estimate based on zbMATH Open (which indexes essentially all mathematical literature since 1868), all of humanity's math output amounts to only 50-100 billion tokens — this run exceeded it.
- Since conventional mathematics rests on just 10 axioms plus a few deduction rules, much of it can in principle be brute-forced — impossibly hard for humans, but computers can do it at a completely different scale.
- The result is therefore like Google announcing it processed X petabytes of DNA data: potentially important for science, but not informative about general intelligence.
- Progress on IQ tests, ARC-AGI, or chess is more revealing, since OpenAI wouldn't burn enormous resources brute-forcing those.
The author concedes tackling such problems is worthwhile and could genuinely advance science — it just says little about how close we are to AGI.
Related event: OpenAI's Costly Navier-Stokes Run Sparks Scooping and Brute-Force Debate(26 posts)→
More from AGI Musings
- Anthropic's S-1 Shows $6.6M Revenue Per Employee as AI Agents Log 6x Human Hours — jacalulu · 2026-10-26
- Ben Recht: The world needs a truly open language model, and it's easier to build than you think — beenwrekt · 2026-10-26
- Which jobs fall to AI first? The rule: any layer whose rules can be written as code — FinanceYF5 · 2026-10-26
- Anthropic S-1: expert-level AI 'within a few years', 90% of code written by Claude — kimmonismus · 2026-10-26
- The identity crisis in software engineering: multiagent systems demand a new problem set — rakyll · 2026-10-26
- Frontier LLMs negotiate contracts poorly and frequently renege for selfish gain, study finds — xuanalogue · 2026-10-26