What is a Context Window?
A context window is the maximum amount of text a language model can hold at once, covering the prompt, any retrieved sources and the answer it generates.
Published
Definition
A context window is the maximum amount of text, measured in tokens, that a language model can hold in working memory at one time. It covers everything at once: the user’s prompt, any retrieved documents, the conversation so far, and the answer being generated.
Why It Matters for AI Visibility
When an assistant answers a question, your page is not alone in that window. It competes for space with several other sources, the conversation history and the system instructions. Only a portion of your content makes it in, and it is rarely the portion you would have chosen.
Two things follow. First, front-loading works: a page that states its main claim in the first hundred words gives the retriever something usable no matter where it truncates. Second, padding is expensive — a long preamble before the substance means the substance may never enter the window at all.
Example
A retrieval system pulling five sources at roughly 800 tokens each spends 4,000 tokens on sources alone. If your relevant answer sits 1,500 words into a page, the chunk containing it has to earn its place on its own merit — the surrounding argument is not coming with it.