It is extremely rare for mathematical problems in the age of modern research math to be open for more than a thousand years. The most famous example is whether there are odd perfect numbers, and that's... about it? Fwiw I am not a qualified historian of math.

Therefore, this question is mostly whether intelligent forms interested in the Millennium problems will still exist by year 3000, and how long they will be able to maintain work comparable to the current age.

I would not go below a base rate of ~85 %.