Advanced – AI & ML Testing
LLM as Judge – ISTQB Definition
Last Updated: July 12, 2026 · SoftwareTestPilot QA Team
Official ISTQB Definition
An evaluation technique using an LLM to score other model outputs on defined criteria such as helpfulness, factuality, or tone.
In simple words
Using a language model to grade the answers of another language model.
Exam tip
LLM-as-judge scales evaluation but must be validated against human labels to avoid bias.
Related terms
Practice this term in the ISTQB Mock Test
40 CTFL v4.0 questions, 60-minute timer, instant scoring and chapter-wise breakdown.
Practice in Mock Test