Comment by pxc

5 months ago

Some LLMs do better than others, but this still sometimes trips up even "frontier" non-reasoning models. People were showing this on this very forum with GPT-5 in the past couple days.

0 comments