AI achieves its first International Mathematical Olympiad (IMO) at least bronze medal in the calendar year X. AI achieves its first IMO perfect score in the calendar year Y. This market resolves to Y-X.

In case that some AI is not explicitly tested on IMO, but e.g. it solves a Millennium problem and there is a widespread agreement that it could get IMO perfect score were it tested, this question could be resolved before an explicit IMO test. In case of such a possibly contentious resolution I plan to poll Manifold Moderators and/or experts in the field whether they would agree with that resolution.