AI glossary · Beginner · Language models
Context window
A context window is the maximum amount of text an AI model can read, keep in mind and work with at one time. It is usually expressed as a number of tokens.
About this term
- Level
- Beginner · Core ideas for first-time readers
- Category
- Basics
- Fields
- Language models
- Added
- 2026-10-07
How it works
The user's question, previous conversation, attached documents, and even the answer the model writes must all fit inside the context window. If this limit is exceeded, older content may be cut off or the request may be rejected altogether. The model refers directly only to what is inside this window when producing an answer.
What shares a single window
Context window (token limit)
InstructionsPrevious conversationAttached documentsCurrent questionAnswer written by the modelOverflow: cut off or rejected
Why it matters
With a large context window, you can put in a long report, a contract or an entire codebase at once and ask questions about it. In long conversations, the model can keep going without forgetting earlier parts. That is why context window size is often presented as a key feature when a new model is released.
Good to know
A large window does not mean the model uses everything in it equally well. It sometimes misses information in the middle of a long text. Also, the more you put in, the higher the processing cost and time. This is why methods such as RAG, which put in only the needed parts, are used alongside it.
Example
When you upload a long PDF to a document summarization service, you may see a message saying 'The document is too long' or get a summary of only part of it. This is often because the document exceeds the model's context window.