Context is all the information available to a model when generating a response — including the prompt, conversation history, retrieved documents, and system instructions. A model's context window (measured in tokens) sets the hard limit on how much information it can consider at once. Longer context windows enable processing entire documents or extended conversations but increase computational cost quadratically in standard attention. Techniques like sliding window attention, retrieval-augmented generation, and context compression help models work effectively within and beyond their native context limits.