SoftwareTestPilot

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