Skip to main content
MANIFOLD
When will the first “Major advance” problem in FrontierMath: Open Problems be solved?
1
Ṁ175Ṁ250
2030
December 2, 2027
24%
Before Jan 2027
34%
Before Jul 2027
66%
Before Jan 2028
76%
Before Jul 2028
87%
Before Jan 2029
90%
Before Jul 2029
96%
Before Jan 2030

This market predicts when the first currently listed problem in the “Major advance” category of Epoch AI’s FrontierMath: Open Problems benchmark will be officially recognized as solved.

Only problems listed in the “Major advance” category at the time this market is created are eligible. Problems added to that category later do not qualify.

Each threshold resolves YES if a qualifying solution is publicly recognized before the date shown, and NO otherwise.

For example, if the first qualifying solution is announced on March 10, 2028:

  • “Before January 1, 2028” resolves NO.

  • “Before July 1, 2028” resolves YES.

  • Every later threshold resolves YES.

The relevant date is the date Epoch first publicly announces or records an eligible problem as solved, not necessarily the date the solution was originally produced.

A solution qualifies when Epoch officially marks the problem as solved or publicly confirms that the solution satisfies its verification process.

Journal publication, peer review, and independent replication are not required unless Epoch makes them a condition for recognizing the problem as solved.

If Epoch withdraws or corrects an announcement before the affected thresholds are resolved, that announcement does not qualify. Once a threshold has been resolved based on Epoch’s official recognition, it will not be reopened because of a later retraction.

Epoch AI’s official FrontierMath website, benchmark records, and public announcements are the primary resolution sources. Official corrections published before resolution take precedence over third-party reporting.

Market context
Get
Ṁ1,000
to start trading!