Poor performance of GPT 4, and confusing description of the Models (GPT 4 and GPT4o)

Performance of GPT-4 is ‘watered-down’, I can say this from the perspective of someone who has used the GPT-4 almost every day for complex tasks since the early launch of GPT-4. Especially in engineering and coding. 1) it is slower in generating answers, 2) it is lazier, and avoids writing scripts, 3) makes very simple mistakes.
Currently, on the model selector of ChatGPT, it says the following: ‘GPT-4: Advanced model for complex tasks’ and ‘GPT4o: Newest and most advanced model’. ‘Most advanced’ I assume meaning the smartest between them but I’m not sure how OpenAI defines ‘Advanced’, because the description for GPT-4 says ‘Advanced model for complex tasks’, so what is the ‘complex’ here? Should I use your most advanced model or the advanced model for the complex tasks? This seems like a wordplay to me and is designed for marketing purposes. I care about the results it produces and I can tell that it is worsened. I don’t have a benchmark on my hand to prove it but I can tell from the experience.