The amount of text an AI model can consider at once when generating responses, measured in tokens. Critical for understanding model limitations.