Skip to content
Superscout
Get Pro
Glossary

Evals

Evals, short for evaluations, are structured tests used to measure whether an AI model, prompt, agent, or LLM application performs well enough for a specific use case.

Accurate as of Jun 4, 2026

Ai TechnicalMetrics Kpis

Shared glossary categories

Related terms

All 185 terms
Inference costInference cost is the per-use cost of serving model predictions or generations in production.Agentic AIAgentic AI refers to AI systems that can pursue a goal by planning, making decisions, using tools, and executing multi-step workflows with some degree of independence.AI wrapperAn AI wrapper is a software product built on top of an existing AI model, model API, or foundation-model provider rather than by training or owning the core model itself.

Superscout maintains this definition in its canonical glossary record and shows uncertainty when meaning varies by context.

Suggest a correctionHow the research works
Sources and known limits

Last reviewed .

This definition does not yet carry a public source link.

Usage can vary across firms, programs, and jurisdictions; this definition describes the venture-scouting context.