Practice · Established · Intermediate

Chain-of-Thought Prompting

Also known as: CoT

Prompting a model to work through intermediate steps before giving a final answer, which improves accuracy on reasoning tasks.

What Chain-of-Thought Prompting is

Rather than jumping to a conclusion, the model is asked to lay out the reasoning path. Spending more tokens on the problem measurably improves multi-step arithmetic, logic and planning.

How it works

Either instruct the model to reason step by step, or supply examples that demonstrate the reasoning. Self-consistency samples several chains and takes the majority answer. Reasoning-tuned models perform this internally and may hide the trace.

Why it matters

It was the discovery that unlocked reliable multi-step problem solving from language models and led directly to dedicated reasoning models.

Common uses

→Maths and logic problems
→Multi-step planning
→Diagnostic and troubleshooting flows
→Complex data transformations

Strengths

✓Substantial accuracy gains on reasoning
✓Makes errors visible and reviewable

Watch for

✓More tokens, cost and latency
✓Stated reasoning may not reflect the real computation

Continue exploring

Reasoning Model Self-Consistency Tree of Thought Test-Time Compute