AI use case

Compare how much code AI coding models add

An open-source benchmark reuses published patches from coding benchmarks and counts the syntax elements each model adds, so you can see which models solve tasks with the least extra code to maintain.