Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
Authors: Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed Chi, Quoc Le, Denny Zhou
Published January 28, 2022
Read the paper: arXiv 2201.11903 · PDF
Why it matters
The practical foundation under every reasoning-heavy LLM deployment today. Even today's reasoning models owe their internal thinking behavior to the CoT family of techniques.
For practitioners, the lesson isn't "add let's think step by step". It's that explicit intermediate reasoning can be engineered, measured, and improved, and that large non-reasoning models benefit from it.
How to use it
If you're using a non-reasoning model, always enable CoT for math, logic, and multi-hop Q&A. If you're using a reasoning model, CoT in the prompt is redundant, because the model is already doing it.
Abstract
We explore how generating a chain of thought -- a series of intermediate reasoning steps -- significantly improves the ability of large language models to perform complex reasoning.
Articles that use this
- When to Use a Reasoning Model, and When Thinking Just Costs MoreTechniques
- Stop Trusting Free-Form Chain-of-Thought. Make Your Model Sign a Logical Certificate InsteadTechniques
- Chain-of-Thought Assumes Left-to-Right. Mercury 2 Doesn't: How to Prompt a Diffusion LLM by Handing It a Skeleton, Not a TrailTechniques
- Adaptive Few-Shot Prompting: Use Semantic Retrieval to Pick Your Examples Instead of Hardcoding ThemTechniques
- Uncertainty Markers Double LLM Reasoning Accuracy: The Self-Correction Prompting Technique You Shouldn't Sleep OnTechniques

