Contributed by the Harvard's AI Safety Student Team community.

0 comments
10 forecasters

What will be the best normalized score achieved on the original 7 RE-Bench tasks by December 31st 2025?

Latest estimate
1.45