AI Humanizer Benchmark

AI Humanizer Benchmark is an independent monthly leaderboard that puts AI humanizers to the test against seven widely used AI detectors. It measures not only whether rewritten text can get past detection, but also how well the output preserves the original meaning and stays natural to read.

Every month, the same texts are run through each tool under identical conditions, so the comparisons remain fair and consistent. Rankings and raw results are published openly, and the scoring rules along with the code are available for anyone who wants to inspect the methodology or reproduce the findings on their own.

The benchmark is operated by the team behind UndetectedGPT, which itself appears in the rankings. To keep that conflict of interest transparent, the site publishes all of its data and methodology, making it possible for readers to review and verify the results independently.