# AI benchmark foundations for buyers

Learn what an AI benchmark measures, how its grader scores a run, what its data licence allows and how to read a reported score before you use it in a decision.

- Level: Foundations
- Length: 40 minutes
- Exam questions: 9, pass mark 80%, up to 3 attempts
- Certificate: valid for 12 months

## What you will learn

- Name the parts of an AI benchmark: tasks, procedure, grader, score and setup.
- Tell the grader types apart: exact, unit-test, simulation, llm-judge, human and mixed.
- Separate the licence of the paper, the code and the task data.
- Read status, saturation, top score and score provenance on a benchmark listing.

## Lessons

1. [What an AI benchmark measures](https://aibenchmarkindex.com/academy/ai-benchmark-foundations/what-a-benchmark-measures.md) (12 minutes)
2. [Graders, data access and licences](https://aibenchmarkindex.com/academy/ai-benchmark-foundations/graders-data-access-and-licences.md) (14 minutes)
3. [Read a benchmark listing](https://aibenchmarkindex.com/academy/ai-benchmark-foundations/read-a-benchmark-listing.md) (14 minutes)

## Questions

### Do I need to know how to code to take this course?

No. The course is for people who read benchmark results to choose or buy a model or tool. It teaches what to check, not how to write an evaluation.

### Does this course rank AI models?

No. It teaches you how to read a benchmark and its scores. It does not name a best model or a best benchmark.

### What does the certificate show?

It is a certificate of completion from this site. It records that you passed this course's exam. It is not a licence, an accreditation or a professional qualification.

Also on the [web page](https://aibenchmarkindex.com/academy/ai-benchmark-foundations).