What is Chain-of-Thought Prompting?

Chain-of-Thought Prompting, explained

Chain-of-thought prompting was described by researchers in 2022, who found that asking models to show intermediate reasoning improved results on arithmetic, logic, and multi-step questions. The idea is simple: instead of asking only for an answer, the prompt asks the model to explain its steps, or includes an example of worked reasoning.

For instance, rather than asking how many delivery vans a bakery needs, a prompt might say: list the number of orders per route, estimate the time per delivery, then calculate the vans required. Many newer models now perform this kind of reasoning internally without being asked. Step-by-step output can make answers easier to check, but the written reasoning is not a guaranteed record of how the model reached its conclusion, so results still need verification.