GPT 5.5 seems to be degraded

After the usage syncing issue fix (and maybe that same week before that fix), I see noticeable degradation in prompt response quality. I know this is subjective and I don’t have solid empirical evidence to show you but the main pattern I am seeing consistently now is its clear inability to adhere or follow the instructions given and at times it seems to eagerly declare an outcome or a patch but it is incorrect or the code quality is bad enough to cause major regressions. My prior impression of GPT 5.5 was that it would be able to fix issues very well. I don’t think the codebase has changed or my workflow has been altered. I am primarily working with Flutter nad Xcode and it seems that it has gotten suddenly incapable of one shotting prompts even for the simplest UI changes which is uncharacteristic of its prior performance. I see several other mentions of this across social media so I can’t be alone in observing these noticeable delta over time. My biggest issue is that using GPT 5.5-med causes the most regressions and coding issues whereas before I never had to reach beyond it to work on UI tasks. Using GPT 5.5-high is also not giving me really noticeable uplift. In desperation I have resorted to using GPT 5.5-xhigh but I notice still that it is not running as long as it used to. Previously xhigh would run for hours and amazing at fixing issues even while I pack them into the prompt. Now it does not seem as capable. I am also observing some changes to ChatGPT Pro models as well. Previously it would think for very long period of time and rarely would it respond quicker than 10 minutes but now I see that it is responding much more quicker and often with less material attached. Hoping these issues can get some attention and investigation. Thanks again for the amazing work OpenAI!