[TeMPOraL]

That's what the "reasoning" models do, effectively. Some LLM services hide or summarize that part for you, other return it verbatim, and ofc. you get the full thing if you're using a local reasoning model.