Comment by CSMastermind

Worth mentioning, though, that people have already tried running all of them through LLMs at this point.

So this is proof of the models actually getting stronger (previous generations of LLMs were unable to solve this one).