Skip to main content
MANIFOLD
How many Erdős problems will be marked solved on erdosproblems.com at the end of 2026?
10
Ṁ1.1kṀ566
Dec 31
27%
≤599
21%
600–629
38%
630–659
6%
660–699
9%
700+

Context:

erdosproblems.com currently displays 1217 problems, of which 565 (46%) are solved.

Tao announced in January that problem #728 was solved more or less autonomously by AI — the first such milestone, then OpenAI's May 20 counterexample to the 1946 unit-distance conjecture became the first historically significant proof from an AI model, and on August 1 Astra added 10 more advances including three additional Erdős problems. Labs have explicitly realized the site works as a benchmark, which means the remaining open problems are now targets in a race. The scoreboard is public: erdosproblems.com

Resolution terms:

• Resolves per the “X problems have been solved” counter on the erdosproblems.com homepage, using the last Wayback Machine snapshot dated on or before Dec 31, 2026 (or a live check on Jan 1 if no snapshot exists).

• Counts all problems marked SOLVED regardless of whether AI, humans, or hybrid teams did it, and regardless of when the underlying solution occurred — the site records status changes when solutions are verified, not when they happen,  so verification lag is part of what you’re trading.

• Growth in the total problem count (it went from 1,179 in February  to 1,217 now) is irrelevant; only the solved count matters.

• If the site is down or abandoned, fallback: solved-status count in the teorth/erdosproblems GitHub database as of the same date.

Get
Ṁ1,000
to start trading!
Sort by: