challenge_10013_1781221989992

A context window is the maximum amount of text an LLM can “see” at once when generating a response.

It includes the prompt, conversation history, tool outputs, uploaded content, and the model’s own reply. If the conversation exceeds the window, older or less relevant content must be dropped, summarized, or retrieved separately.