Concept
Knowledge Cutoff
The most-recent date represented in a model's training data â beyond which it has no knowledge.
Definition
Knowledge cutoff is the latest date covered by a model's pre-training corpus. Without external tools the model cannot reliably know about events after this date. Frontier models typically lag by 6â12 months; tool-using agents fill the gap with web search.
Common use cases
- Application disclaimers
- RAG necessity
- Date sensitivity