ReAct (reasoning and acting)
ReAct (reasoning and acting) is a prompting pattern in which a language model alternates between writing a reasoning step, taking an action such as a search or tool call, and reading the result before its next step.
Also called: ReAct pattern
By Divya Sharma · Checked Nov 15, 2026 · How we review
ReAct was introduced in a 2022 paper by Yao and colleagues. Its idea is simple: let the model think out loud and act in the same transcript, so each thought can draw on what the last action returned. It became the template for most tool-using agents that followed.
How it works
The prompt asks the model to produce turns in a fixed shape. A Thought line says what it needs to find out. An Action line names a tool and its input, for example a search query. The program runs the tool and appends an Observation with the result. The model then writes the next Thought, and the cycle repeats until it gives a final answer. Today this usually runs through native function calling instead of parsing text, but the shape is the same agent loop.
ReAct versus chain of thought
Chain of thought has the model reason step by step using only what it already knows. ReAct adds actions, so the model can fetch facts it lacks and correct course when a result surprises it. That makes it less prone to invented facts on questions that need outside information.
Limits
Each step is a model call, so long tasks get slow and costly, and a bad early thought can send the loop in circles. Agent frameworks such as LangGraph add step limits and state handling around the basic pattern. The name has nothing to do with React, the JavaScript library.
Related terms
- Agent loop
- An agent loop is the repeating cycle at the heart of an AI agent: the model asks for a tool call, the application runs it and returns the result, and the model decides what to do next.
- Chain-of-thought
- Chain-of-thought (CoT) is a prompting method in which a language model writes out intermediate reasoning steps before its final answer, which tends to improve results on maths, logic and other multi-step problems.
- Tool calling
- Tool calling, also called function calling, is a model API feature where you describe functions with a schema, the model replies with a structured request to call one, and your code runs it and returns the result.
- AI agent
- An AI agent is a system in which a language model decides its own next steps: it calls tools, reads the results and repeats until the task is done, instead of following a fixed script.
- Reasoning model
- A reasoning model is a language model trained to work through a problem in a stream of intermediate thinking tokens before giving its final answer, which tends to help on maths, coding and multi-step tasks.