GenGrades is an independent review site that grades generative AI tools and models with standardized tests: the same prompts and inputs for every model, real outputs published with the review, and scores on a 0 to 10 scale with public weights. Tools are grouped by job — chat and agents, image, video, audio, and API aggregators — and each category has its own test set, rubric, ranking, and published weights. Outputs are scored blind where possible; a language model may pre-screen, but a person sets the final score. New model versions get a new, dated run. Reviews list price, limits, and who should not pay, with the source and the day those facts were checked. Grades are not for sale. As of 11 November 2026 the site had 98 tools and 16 models graded. Reading the reviews is free.