Directory
Search 264 AI benchmarks by category, with data sources and licences for each.
The index catalogs 274 AI benchmarks with source-linked facts. It warns users to check the cited primary source before using a result. It recommends specific benchmarks such as AdvBench for safety.
It lists the number of benchmarks per category, for example, Multimodal (51). It provides guides on how to choose and run benchmarks. It includes a guide on saturated benchmarks and how to spot them.
It explains where benchmark data comes from and what licences allow. The data sources for benchmarks are provided under the CC0-1.0 license.
Summary drafted from the project's own website. Every sentence is backed by text on that page and was reviewed before publishing.