Context Window
The maximum amount of text — measured in tokens — a language model can consider at once, covering both your prompt and its reply. Go past it and the earliest tokens fall out of view.
Think of It Like This
It's the model's desk space — only so much paper fits on top at one time.
Everything the model reasons over has to fit inside the context window: system instructions, chat history, pasted documents, and the answer it's still generating. Windows have grown from a few thousand tokens to hundreds of thousands, which is what makes "just paste the whole document" workflows possible. The catch is that self-attention cost climbs quickly with length, so a bigger window is never free.