FrontierThe story, in brief
OpenAI claims it solved an 80-year-old math problem — for real this time
OpenAI's reasoning model just did what mathematicians couldn't for 80 years. The skeptics are now believers.
The KeyNews take
Why it matters
A major capability milestone for AI reasoning models—solving a formally verified geometry conjecture—signals that advanced reasoning is moving from benchmark claims to real mathematical discovery. This validates the reasoning-focused model architecture that's reshaping the competitive landscape.
The key facts
4 to knowOpenAI reasoning model disproved Erdős–Ko-Rado conjecture in geometry (unsolved since 1946)
Independent mathematician verification provides credibility after prior false claims
80-year-old unsolved problem solved by AI reasoning capability
Published May 20, 2026 on TechCrunch
Go to the source
TechCrunch AItechcrunch.com
Publisher excerpt: OpenAI claims its reasoning model disproved a geometry conjecture unsolved since 1946 — and this time, the mathematicians who exposed its last embarrassing claim are backing it up.