Evals

Evals use an LLM judge to score a sample of your GenAI spans, so the quality of your agents' answers sits next to their latency and cost.

Evals are in beta. Each eval belongs to one dataset and returns a numeric score, a boolean, or a categorical label for every span it judges. For the full terminology, see the Evals Key Concepts.

Further Reading

• About Evals — What evals do, where to find them, and their limitations.
• GenAI Spans — Send the spans and attributes evals depend on.
• Create an Eval — Define which spans to judge, the judge model, and the output.
• Key Concepts — Terminology for triggers, sampling, verdicts, and credits.