I put GPT-5.2 through an 13-round test, and the AI model raised some serious questions

After 10 text and 4 image tests, OpenAI’s latest model barely beats GPT-5.1. What are Plus subscribers really getting?