OpenAI claims ~10,000 agents solved Navier–Stokes in 88 hours
An unreleased model is said to have resolved the Navier–Stokes existence and smoothness problem. The result is unpublished, unverified and disputed — and the Clay Institute still lists it as 'active'.
OpenAI announced that roughly 10,000 AI agents running on an unreleased model — more powerful than GPT-6 Astra — had solved the Navier–Stokes existence and smoothness problem in 88 hours, concluding the governing equations can produce sudden infinities.
The claim arrived shadowed by a dispute: NYU’s Tristan Buckmaster said he and Anthropic researcher Levent Alpöge had been working the same problem, and that OpenAI had moved to publish after learning of their progress. The Clay Mathematics Institute still lists the problem as active rather than solved, and under its rules a prize cannot be awarded until at least two years after publication.
Why this is flagged unverified
This entry carries verified: false, so its weights count at half strength in the index.
That is deliberate: a corporate capability claim that is unpublished, unverified and actively
disputed should not move the ARQ as much as a confirmed event. Without that damping, hype would
inflate the index.
Risk 4 · Mitigation 1, at 50% confidence.