Home › Glossary › Benchmark
Evaluation · beginner

What is Benchmark?

A plain-English explanation of Benchmark (AI Benchmark) — what it means, why it matters, and how it is used in AI.

Benchmark
AI Benchmark
A benchmark is a standardised test or dataset used to measure and compare AI model performance.
"When a new AI model is released, labs publish benchmark scores so people can compare it to existing models."

Also known as: Evaluation benchmark, AI benchmark, model evaluation

EvaluationMMLUAccuracy

Why does Benchmark matter?

Benchmarks are used to compare models, track AI progress over time, and choose the right model for a specific task.

Practice this term

The best way to remember Benchmark is to practice unscrambling it. AI Terminology Scrambler uses spaced repetition to help you learn and retain AI vocabulary in just a few minutes a day.

Practice Benchmark now →

Related AI terms

EvaluationMMLUAccuracy