Context window: The context window is the maximum amount of text, measured in tokens, that a language model can consider at one time. It includes your prompt, any pasted documents, the conversation so far, and the model's reply.
Think of the context window as the model's working memory. Everything it knows about your specific request has to fit inside it: the instructions, the files you attached, every earlier turn of the conversation, and the answer it is writing. Nothing outside the window exists to the model.
Windows have grown a lot. Early chat models handled a few thousand tokens, roughly a short article. As of mid-2026, leading models accept hundreds of thousands of tokens, and some accept a million or more, enough for several books. Check the vendor's documentation, because limits differ by model and plan.
Bigger is not the same as perfect. When a conversation runs long, older messages may be trimmed or summarized, which is why an assistant sometimes forgets a rule you set an hour ago. Models can also pay less attention to material buried in the middle of a very long input. Put the most important instructions and facts near the start or the end, and restate them when a thread runs long.
Example at work
A consultant loads twelve client interview transcripts into a single chat and asks for common themes. The first few answers are sharp, but after twenty follow-up questions the model starts contradicting earlier findings. She starts a fresh chat, pastes her saved summary of the themes, and continues from there.
Why it matters
The context window sets the practical ceiling on what you can hand an AI tool in one go. Understanding it tells you when to split a task, when to start a new conversation, and why restating key instructions late in a long thread gets better results.
Related terms
- TokenA token is the unit of text a language model reads and writes, usually a word or a piece of a word. Models measure input limits, output limits, and pricing in tokens; as a rough rule, 1,000 tokens is about 750 words of English.
- Large language model (LLM)A large language model (LLM) is a neural network trained on enormous amounts of text to predict the next word, then tuned to follow instructions and hold a conversation. LLMs power ChatGPT, Claude, Gemini, and Microsoft Copilot.
- PromptA prompt is the text (and sometimes files or images) you give an AI model to tell it what you want. A good prompt states the role the AI should play, the task, the relevant context, the output format, and any constraints.
- Retrieval-augmented generation (RAG)Retrieval-augmented generation (RAG) is a technique in which an AI system first searches a set of documents for relevant passages and then gives those passages to a language model to answer from. It lets an assistant use your own, current information.
- SummarizationSummarization is the use of AI to condense a long document, transcript, thread, or dataset into a shorter version that keeps the key points. It is one of the most reliable everyday uses of language models, with known failure modes worth checking.