Tokens & context windows

Language models process text as tokens — roughly 3/4 of a word each. Every model has a context window: the maximum number of tokens it can process in a single request.

This matters for prompt design because everything in your prompt — instructions, examples, and the user's input — must fit within the context window. Longer contexts cost more and can dilute the model's attention.

Key takeaway: Be concise. Put the most important instructions first. Use delimiters to separate sections clearly.