Kind of wild that we're finally getting benchmarks for AI-generated API testing. Feels like the equivalent of SWE-bench, but for finding actual bugs instead of

Kind of wild that we're finally getting benchmarks for AI-generated API testing. Feels like the equivalent of SWE-bench, but for finding actual bugs instead of writing code.