Chain of thought

Chain of thought started as a prompting trick: asking a model to think through a problem step by step before answering. Writing out intermediate steps gives the model room to work, and it often catches errors it would otherwise make by jumping straight to a conclusion.

Modern reasoning models do this natively. Features such as extended thinking let the model produce a private reasoning section before its final reply, and you can often control how much thinking it is allowed to do. More thinking usually means better answers on hard problems, at the cost of more tokens and more time.

Treat visible reasoning as a helpful signal, not a guaranteed record of how the model reached its answer. It is useful for debugging, but you still need to verify the final output.