That's still selective compared to the volume of applicants though and the output isn't universally stellar but generally regarded as pretty high. Compared to YC's recent batches with a large number of ChatGPT and other gen AI model wrappers.