# Which benchmark should I use?

Choose the capability, maximum acceptable saturation, and data access you need to narrow the benchmark list.

Use the tool on the [web page](https://aibenchmarkindex.com/tools/which-benchmark-should-i-use). It works without JavaScript.

## How to use it

1. Choose capability to evaluate.
2. Enter maximum saturation.
3. Choose data access needed.
4. Read the result.

## How it works

The tool counts published listings in the chosen category, then counts those whose recorded saturation is no higher than your limit. Missing saturation values do not enter the second count.

The result tells you how to form a shortlist. It does not declare one benchmark best: task fit and grader validity matter more than a single ranking number. Data-access advice is applied after the saturation count because the current directory tool engine cannot filter an enum attribute in that count.

> The index is a starting point, not proof that a benchmark is valid for your decision. Read the benchmark's paper and licence.

## Questions

### Is a lower saturation always better?

No. It gives a benchmark more headroom, but only if the tasks and grader measure the capability you care about.

### What does open data mean here?

The listing says the task data can be obtained without an application or private request. Always read the recorded licence before reuse.

> **Check before you compare.** Benchmark methods and leaderboards change. Follow the cited primary source before using a result in a decision.