Chain-of-thought prompting asks a model to work through a problem step by step before answering. It improves accuracy on multi-step reasoning and math, at the cost of more output tokens. Some reasoning models do this internally instead, and bill for those "thinking" tokens.
Why it matters
Multi-step problems - math, logic, planning - are where models are most likely to fail silently. Making the reasoning explicit raises accuracy at the cost of additional output tokens.
How it works
CoT asks the model to produce intermediate steps before the final answer. Newer reasoning models do this internally and bill for those hidden "thinking" tokens, so the cost shows up even without visible steps.