Explained · 2 min
What is a context window?
How much a model can read at once, and why it runs out.
Written by AI and checked against our editorial standards. Not reviewed by a human editor.
The context window is how much text a model can look at in one go. It includes your question, any documents you paste in, the conversation so far and the model’s own answer.
It is measured in tokens. A token is a small piece of text, roughly three quarters of an English word. A window of 1 million tokens holds about 750,000 words, or several thick books.
Think of it like this: the context window is the model’s desk. Everything on the desk it can read and use. Anything that does not fit on the desk is out of sight.
When a conversation gets longer than the window, something has to go. Tools like Claude Code and ChatGPT summarise older parts of the chat to make room. That is why a model can “forget” details from early in a long session.
A bigger window is not automatically better. Long inputs cost more, because you pay per token. Models can also be less precise about details buried in the middle of a very long text.
Why it matters to you: for big documents or codebases, check the window size. For everyday questions, a shorter and well-chosen context often gives better and cheaper answers.
Learn more: Anthropic docs: Context windows