---
title: "Context window"
description: "The context window is the maximum amount of text, measured in tokens, that an LLM can take in at once: the question, the instructions, and all supplied data together."
canonical: "https://www.whatsnext-ai.com/glossary/context-window"
---

# Context window

The context window is the maximum amount of text, measured in [tokens](https://www.whatsnext-ai.com/glossary/tokens), that an [LLM](https://www.whatsnext-ai.com/glossary/llm) can take in at once: the question, the instructions, and all supplied data together.

## Definition

If the conversation or supplied document is larger than the window, the oldest information drops out of view and the model "forgets" it. That is why agents in daily use rely on [RAG](https://www.whatsnext-ai.com/glossary/rag): they fetch only the pieces relevant to each question instead of stuffing everything into the window. More on how agents work: [What is an AI agent?](https://www.whatsnext-ai.com/blog/what-is-an-ai-agent)

## Related

- [Custom AI Development](https://www.whatsnext-ai.com/custom-ai-development)