What is a context window in AI and why does it matter?

A context window is the amount of text an AI model can 'remember' and consider at one time during a conversation, measured in tokens (roughly 750 words per 1,000 tokens). When you chat with an AI, everything you've written plus everything it has written back sits inside this window. Once the conversation exceeds the limit, the oldest parts get dropped, and the AI literally forgets them. Here's a concrete example: if a model has a 128,000-token context window, you could paste in a 200-page document and ask questions about it. But if your conversation with the AI goes on long enough, those early pages will fall out of the window, and the AI will start answering as if they never existed. I've run into this when working on long editing sessions. The AI suddenly 'forgets' a character's name or a rule I established twenty messages ago. That's not a bug — it's the context window filling up. A practical tip: for long projects, periodically summarize key points and start a fresh chat with that summary as your opening message. It resets the window while preserving what matters. Most major AI tools now offer windows between 32,000 and 1 million tokens, with larger windows costing more to run. For a deeper dive on how context affects content creation, see our guide on AI content workflows (/blog/ai-content-creation-workflow). **Related**: How many tokens is a typical conversation? | Do longer context windows make AI smarter?