There are a lot of new techniques to "frack" LLMs to squeeze more out of them.

If you get just a single english-language append-only log designed for a human, you make the LLM take its rich understanding and distill it out through a teeny straw of a single, human-understandable line of thought.

Techniques like test-time compute in O1 and similar models allow the model to spray out lots of low quality ideas and then refine.