Knowledge cutoff
The point after which information was not included in a model's training data.
Illustrative conversation
- Can this model know which library version shipped yesterday?
- Not from its training cutoff alone; it needs a current registry, release note, or web source in context.
Definition
A knowledge cutoff is the latest date a provider associates with the information represented in a model's training data. It is a boundary on recency, not a promise that every earlier fact is present or correct. Different models and snapshots can have different cutoffs, so questions about later events should be grounded with current tools or sources rather than answered from parametric knowledge alone.
Reviewed
Source