Lesson 4 of 87 min
Learning goal: Understand tokens and the context window — capacity, limits and costs.
Models process tokens rather than words exactly as people do. Token counts affect limits, processing and some API costs. The context window contains instructions, prompts, files, history, tool results and output available for a request. More capacity does not guarantee perfect recall, and context is not permanent memory.
Retrieve relevant sections from several long reports before comparing them.
Assuming longer chats always help.
Restructure a request as Goal, Context, Requirements and Output.
Curated workflow shortcuts from the Academy library — not necessarily native Claude commands.
Does a large window guarantee every detail is used?
Answer: No.