alt Hacker NewsThey're certainly aware of the test, but a turtle doing a kickflip on a skateboard? I seriously doubt they train their models for that.
https://x.com/JeffDean/status/2024525132266688757
If anything, the disastrous Opus4.7 pelican shows us they don't pelicanmaxx
I think I found the leaked Claude Mythos version of the turtle benchmark: https://www.youtube.com/watch?v=l82XWTKLZuk