Chain of Thought

What is Chain of Thought

Chain of thought is a model working through intermediate reasoning steps on the way to an answer, rather than jumping directly from the question to a final conclusion. Instead of guessing an answer in one leap, the model lays out the pieces: what needs to be true, what follows from what, and arrives at the conclusion as the last step of a visible chain.

This tends to genuinely help on multi-step problems, where an error anywhere in a direct jump to the answer is invisible, but an error in an explicit chain of steps is often something the model catches itself, or that a human reviewer can spot and correct. It is less useful on simple, single-step questions, where the extra steps add length without adding accuracy.

Chain of thought is related to reasoning effort but not identical to it, reasoning effort is a setting controlling how much internal deliberation happens, chain of thought describes the shape that deliberation takes when it's made explicit. Higher reasoning effort often produces a longer, more thorough chain, but the two concepts describe different things: one is a dial, the other is a pattern.