Groundedness measures how well an answer is supported by the provided context. It is the RAG counterpart to hallucination: a grounded answer cites or follows from the retrieved passages, while an ungrounded one introduces facts the context never contained. Evaluate it reference-free against the context.

Why it matters

Groundedness is the RAG-specific quality bar: does the answer actually follow from the context the system retrieved? It is the main metric for trusting RAG output.

How it works

An answer is grounded when its claims are supported by the provided context. It can be scored reference-free against the context, making it a natural evaluator for production RAG traffic.

Example

An answer that states a refund window the retrieved policy never mentions is ungrounded, even if the sentence reads perfectly.

← Back to the full glossary

Put the platform behind the terms

Route, evaluate, and monitor every AI request from one OpenAI-compatible platform.

Start Free → Explore the Features