What is Chain-of-thought prompting?
Also called: CoT · think step by step
Chain-of-thought prompting asks the model to reason through a problem in steps before giving the answer, which measurably improves accuracy on maths, logic and multi-part tasks.
Because a model generates one token at a time, letting it "write its working" gives it room to compute. "Think step by step, then answer" turns a one-shot guess into a sequence of smaller, checkable steps. On reasoning-heavy tasks the gains are large; on simple lookups or creative writing it mostly adds length.
Newer "reasoning" models do this internally, but explicit steps still help when you want to see the logic, catch a wrong assumption, or constrain how the model approaches the problem.
Before answering, list the assumptions, work through the calculation step by step, check the result for order-of-magnitude sanity, then give the final answer on its own line.
- Use it for reasoning tasks; skip it for short factual or creative ones.
- Ask for the final answer separated from the reasoning so it's easy to use.
- Pair it with self-critique on high-stakes work.
Use it right now
Ask our brain anything on the homepage — it remembers the whole conversation — or write a brief in the Studio and see the prompt it compiles to.