Context Window
The amount of text — measured in tokens — that a language model can consider in a single call.
Definition
The amount of text — measured in tokens — that a language model can consider in a single call. Everything the model knows about the current task lives inside this window: instructions, prior conversation, retrieved documents, tool results. Run out of context and earlier information starts dropping out.
Example
Pasting a 500-page contract into a model with a 200K-token window works; pasting it into a 16K-token window does not. Managing what goes into the window is one of the core skills of operating AI systems.
See it in context Learn how Context Window fits into the bigger picture of how software actually works.
Read the Guide →