There was a good comment on the Pelican bicycle svg yesterday about how these models aren't getting much better beyond what the companies focus training them on. I think that's what's happening in this case too, they probably put this in the training set.
- Claude isn't doing that
as evidence to support the assumption that
- it's a marketing trick
Which is obviously non sequitur, as if it were a marketing trick, Anthropic could do it too. Anthropic isn't known for not spending on marketing.
Honestly, nowadays I question human's reasoning ability more than I question AI's.
Because Claude can't do it. Anyone who tells you that Fable is better than GPT 5.6 at pure math is lying to you.
You are correct that LLMs are trained on existing proofs but hiring researchers to solve unsolved problems is just unrealistic, both in terms of how none of the mathematicians simply came out and took credit for their own discovery or exposed this, and how training sets are not easily memorized (rather, the meta techniques are learned).
OpenAI just has better training methods and techniques for pure math over Anthropic, it’s one of their biggest strengths
I hope people are screenshotting this stuff. This really needs to be documented. It's remarkable how wild it's getting.