Imagine you're having a conversation with someone who can only remember the last N words of everything that's been said. Earlier parts of the conversation are simply gone for them. That's exactly how AI works. Every model has a context window — a hard limit on how much text it can 'see' at once, measured in tokens (roughly ¾ of a word each). When the conversation exceeds this limit, the oldest messages fall off the edge and the AI can no longer reference them. Select a model below, then add messages to watch the window fill up — and see what gets forgotten.
Context used: 0 / 4,096 tokens (0%)
Why This Matters
Every time you start a new chat with an AI, you're starting with a fresh, empty context window. As the conversation grows, older messages get pushed out — the AI literally cannot see them anymore. This is why AI assistants sometimes seem to "forget" things you mentioned earlier, why very long documents need to be summarized before an AI can process them, and why some AI services charge more for longer conversations. Larger context windows (like Claude's 200K tokens — roughly 150,000 words!) are one of the most valuable improvements happening in AI right now.