The second batch of “First Proof” problems is meant to evaluate AI’s usefulness for research-level math. The best model got six or seven of the ten questions right.
Results that may be inaccessible to you are currently showing.
Hide inaccessible results