04 · How Models Run
Context Window
How much a model can “see” at once
The max text (in tokens) a model holds at once — your prompt, its response, and any pasted documents. Bigger window = considers more (e.g., a whole book). Exceed it and the earliest content falls out of view.
Concrete example
Paste a 300-page contract into a large-context model and it can answer about page 290; a small-context one will have forgotten the start.
Why it matters
Sets the practical ceiling on how much material a model can reason over in one go.