Context Window vs Knowledge Cutoff
People often think a model with a big context window knows recent news. They are two different limits. The knowledge cutoff is the date its training text stops, so it is about what the model learned. The context window is how much text it can read right now, so it is about what you hand it. Pasting a fresh article into the window is how you get past the cutoff.
Context Window
The most text, in tokens, a model can read at one time.
The context window is the most text a language model can take in at one time. It is counted in tokens. It holds all of one request: the instructions, the chat so far, any pasted files and the answer being written. Sizes run from a few thousand tokens in small models to over a million in some large ones. Text that does not fit is cut off. The model cannot see it.
Knowledge Cutoff
The date a model's training text ends. It knows little after that.
A knowledge cutoff is the date after which a model's training text stops. The model knows little or nothing about events after it. A model is often released months after its cutoff and used for a year or more beyond that. So it may not know about new products, prices, laws or news. Tools that let the model search the web or read your files fill that gap.
In one line
- Context Window
- The most text, in tokens, a model can read at one time.
- Knowledge Cutoff
- The date a model's training text ends. It knows little after that.