OpenAI says it cracked 90-year-old maths problem in 88 hours — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI claimed it solved the Navier-Stokes existence and smoothness problem — a fluid dynamics puzzle lacking a proof for 90 years — in 88 hours using roughly 10,000 AI agents trained on a new internal model.
- The 10,000 bots exchanged nearly 3 million messages and 130 billion output tokens on the problem alone, an effort OpenAI's own pricing pegs at roughly $10M.
- OpenAI's solution resolved 2 of the 4 statements required for the $1M Millennium Prize; the company says it is not claiming the prize, and the proof awaits independent verification by The Clay Mathematics Institute.
- Tristan Buckmaster, a NYU math professor working with Anthropic mathematician Levent Alpöge, alleged OpenAI only began the problem after "information about our progress had been passed to OpenAI," sharing email exchanges as evidence.
- OpenAI congratulated the "concurrent work" but denied seeing it "through any means," conceding it "cannot rule out" that de-identified data from their use of OpenAI's Codex tool improved its models.
- OpenAI said it began training the new model at the end of August, then deployed bots on 1 September after hearing "rumours" that two Millennium Prize problems had been resolved.
Why it matters: OpenAI's $10M proof effort covers only 2 of 4 Millennium Prize statements and awaits independent verification, but the bigger fallout is reputational: a NYU professor's allegation that OpenAI only began the problem after learning of his parallel Codex-assisted work puts the company's training-data sourcing on the defensive.
Ask SkimNews
