Juries of Models Metric

What is the Juries of Models Metric?

The Juries of Models Metric aggregates judgments from multiple large language models on the same output to reduce single-judge bias and variance in LLM-as-judge evaluation.

Use juries for subjective quality; keep deterministic checks for schema, citations, and hard safety rules.

Related Giskard articles

Combine juries with hard checks

Pair multi-judge rubrics with Giskard deterministic gates for format and safety. Learn more.

Further reading

Authoritative reference: Judging LLM-as-a-Judge (Zheng et al.).

Get AI security insights in your inbox