A context window is the amount of input and prior conversation that a language model can consider when generating a response.
A larger context window can support longer documents and richer instructions, but it does not guarantee perfect recall, reasoning, or source fidelity.