The context window is an AI model's working memory: the maximum amount of text — measured in tokens — it can take into account at one time. Everything counts against it: your instructions, the document you pasted, and the conversation so far. Once the limit is reached, the model literally cannot "see" whatever falls outside the window.
This explains some familiar behaviour. A chatbot that "forgets" early parts of a long conversation has pushed them out of its window. A summarizer that misses points from a very long report may only have received part of it. Newer models have much larger windows — hundreds of pages in some cases — but bigger is slower and costlier, so the limit still shapes how AI tools are built. Splitting long documents into sections remains a reliable workaround.
Related terms
Token · Large Language Model (LLM) · Prompt