Anyone else incredibly impressed by Claude Fable? Seems bullish for semis.

For instance, I asked it a question about a chess position, and rather than hallucinate and guess, like prior models would have, it went out and downloaded a chess solver called stockfish, installed it onto its operating system (I'm sure that's not the right terminology but you get the point), ran it, and then explained why stockfish did what it did. The fact that it recognized that it would be a tough question to answer, and went out on its own found and utilized a tool was very impressive to me. I know absolutely 0 about coding and had it create a coaching app for myself, the whole process took about 30 minutes. The hallucination issue is one of the biggest obstacles limiting AI TAM, and I'm seeing more and more self awareness out of these models.