What is Context Window?
Also called Context length
Definition
A context window is the amount of information a model can consider within a generation request, measured in tokens. Depending on the model, that budget includes instructions, conversation history, retrieved material, and generated output.
The operation of Context Window is explained.
The context window is working context rather than the model’s entire training corpus. A system must select what to include when inputs approach its limits. Larger capacity does not ensure that every detail will be recalled accurately.
The relevance of Context Window is explained.
A SaaS assistant comparing security documents may need to select the passages relevant to encryption and retention. Loading unrelated manuals can crowd the context without improving the buyer’s answer.
The evaluation of Context Window is explained.
Context planning budgets for input and output under the model’s documented rules.
Evaluation tests whether qualifications survive retrieval and summarization and measures correctness with representative documents rather than maximum input length alone.
Frequently asked questions about Context Window
Is context the same as training data?
Context is information supplied during a request, while training data shaped the model before that request.
Does larger context guarantee better answers?
A larger context window does not guarantee better recall or factual accuracy.
Do tool results consume context?
Tool results included in the request consume context under the model’s accounting rules.