nostrich on nostr

I think Kimi is better than Claude for complex mathematical reasoning. It's probably more or less the same basic model but they seem to have tuned it to really investigate and reflect more carefully.

Big Barry Bitcoin · 6d

I find it hard to believe Kimi compares to Claude... Maybe it's psychological but I just don't see it when I try it myself. (Coding and reasoning about code)

YODL · 11h

Are you having success yet with the type of crypto stuff you investigate and work on? I tried gpt 5.4 on some Goodstein Theorem papers the other day and was reasonably impressed by its ability to answer some basic questions from papers I fed it, but I could still feel a little of the gaslighting her...