Post #1618938

If you look at the [list of tasks](https://arcprize.org/tasks),, you can see how the 4 frontier models did. Some of them did complete one or two levels of one or two tasks. None of them completed a whole task. Some of the reasoning logs are funny in the replays.