Your question is Context Windows in Long Inputs. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
You are building an LLM feature that has to read long documents, chat history, or multiple retrieved passages before answering. You notice that model quality drops as more input is packed into the prompt, even when the model technically supports the full length.
Explain what a context window is, how tokenization affects it, and what technical challenges appear when processing long-context inputs. What practical techniques would you use to handle those challenges?