Original research

Benchmarks What the engines get right, and what they get wrong.

Measurements of the engines behind the site’s tools, run against fixtures anyone can check and published with the method, the versions, every failure and the files to run them again.

Accessibility engine

Contrast methods

Benchmarks
2
Last run
Rerun files
9

Published

Two benchmarks

Standards

What every benchmark here publishes

  1. Fixtures others wrote. Test material from the W3C, or a sample defined in advance, never cases chosen after seeing the results.

  2. The failures, one by one. Every miss and every false alarm is listed with a link to the case, not just a score.

  3. Versions pinned. The engine, the reference tools, the browser and the fixtures are named with versions and checksums.

  4. A way to run it again. The code, the exact engine that was tested and the raw results are published with each run.

Found a mistake? Write to us and we will correct it.