Context Window
The context window is the maximum amount of text, measured in tokens, that a language model can process at one time – including the input (prompt and documents) and the generated response.
In practice
If a long document or an entire website is handed to a model for analysis, content beyond the context window is usually truncated or ignored silently in chat interfaces; when an API is used directly, the request is instead normally rejected with an error message. This matters particularly for RAG-based AEO strategies, where relevant passages have to be selected deliberately rather than passed over in full. Larger context windows allow more input, but they do not replace careful selection of the most important information.