Is it me of is 5.6 a step down from 5.5?
I have been use ChatGPT since 3.5 and this is the first time I have thought a model was not an improvement over the last. Since its release 5.6 continuously overengineers, gatekeeps, overcomplicates and creates prison like feature releases for everything. For instance, I created an editorial flow for a publication and each step of that editorial flow it added a conditional release gate despite being told not to do so. I have updated AGENTS.md to counter this. I have added statements to promprts and intervened in its work to ask it -Do you think you are overengineering? It then agrees with me and then proceeds to do the same thing. Its lucky that I am running Kimi 3 and Fable 5 otherwise I would be locked in a prison.
And like a chococolate bar, I have noticed that the Pro plan has been shrinking. I have tried using the Sol medium for tasks but it simply ignores me and does the opposite.
Is any one else facing similar issues with 5.6? Have you gone back to 5.5? Have they made 5.5 worse? I am at the point that I might just permanently switch to Kimi 3 and Fable 5.