Every large language model works with a fixed amount of text at a time. This limit is called the context window. It includes everything the model can see in a single exchange: your instructions, any documents you share, earlier messages in the conversation, and the reply it is writing.

Measured in tokens

Context windows are measured in tokens, not words. A token is a small piece of text, often a short word or part of a longer one. When a conversation grows past the limit, the oldest material drops out of view or has to be shortened. The model does not keep it anywhere else. For that exchange, it is simply gone.

Read More