← Back to Glossary
Chain-of-Thought (CoT)
Prompting an LLM to output step-by-step intermediate reasoning out loud before giving its final answer.
Why it exists
Complex math or logic problems fail when forced to answer in one step. Generating reasoning tokens gives the model computational scratchpad space.
How it works
Adding phrases like 'Think step-by-step' forces the model to generate intermediate reasoning tokens prior to outputting the final answer.