Glossary
Faithfulness
On this page
The short answer
- Faithfulness is whether every claim in an output is supported by the context you provided, such as retrieved passages or a knowledge list. An answer can be fluent and still unfaithful if it adds a fact the context does not contain.
- This is a glossary entry, not a product claim or a benchmark result.
- All terms · Templates · Guides.
Definition
Faithfulness is whether every claim in an output is supported by the context you provided, such as retrieved passages or a knowledge list. An answer can be fluent and still unfaithful if it adds a fact the context does not contain.
In practice
Test it with cases that include both a supported fact and a question the context does not answer. The RAG template is built around this. Faithfulness is not the same as world-knowledge factuality: a faithful answer may still be outdated if the context is.
Litmiq does not compute a built-in faithfulness metric. You write expected behavior (“cite [P1]; must not use the conflicting flyer”) and a judge or a person checks it.
Related terms
FAQ
Does Litmiq implement “Faithfulness” as a built-in metric?
Only where the definition itself says so. Many terms are industry vocabulary. Litmiq is a single-turn, BYOK prompt tester with an LLM judge, not a full evaluation platform.
Where is the rest of the glossary?
The hub at /glossary lists every term. This page is the standalone entry for Faithfulness.
See the term in a suite
Many glossary entries have a matching template you can run in the browser.
- Free, no signup
- Runs in your browser
- Your OpenAI key, sent only to OpenAI