Context Window

Context window refers to the maximum number of tokens an LLM can process in a single interaction, including both input prompt and generated output. When exceeded, older parts are truncated. Understanding this constraint is crucial for prompt engineering—you must balance providing sufficient context with staying within token limits.