How much text an AI can "see" at once. The limit that determines whether it can read a tweet or an entire book.
The context window is the total amount of text an AI model can hold in its short-term memory during a single conversation. It includes everything: the system prompt, your messages, the model's replies, and any documents you've pasted in. When you exceed it, the AI starts forgetting the earliest parts.
A small context window means the AI can only handle short, focused conversations. A large one means you can paste an entire book and ask questions across it. This is the difference between an AI that's useful for quick chats and one that's useful for serious work with long documents.
Numbers shift quarterly as models update. The relative ordering tends to stay similar.
Models count text in tokens — chunks roughly ¾ of a word. So a 100,000-token context fits about 75,000 words. The Bible is ~750,000 words for reference.
Bigger context windows aren't always better. Models can technically read 200,000 tokens but their quality at the very end of that window degrades. The "needle in a haystack" tests show even the best models miss details buried deep in long inputs.
Practical advice: paste only what's relevant. Don't dump entire databases when a focused excerpt would work better.