Breaking Azure Support on Purpose every other Update and making it a second class citizen?

[nesdeq]

I get it, OpenAI wants to push direct customer access and their own infra but it is really obvious NOONE is regression testing anything Azure. Not even basic stuff like deployment vs modelname. For older versions Azure only could be made to work AFTER doing OpenAI signin or meddling in config files. Recent 0.4.0. update just broke it in general dropping 40x-s here and there. I am wondering if this is just my anecdotal experience or what other fellow AzureOpenAI users report. Maybe we can have this just as a clean option during setup so it JUST. WORKS.

[supernitin]

Another release breaking Azure support. I'm guessing because Codex is defaulting to gpt-5.1 models which OpenAI chose to launch ahead of making available to Azure. I would think MSFT would be more diligent in their terms with OpenAI to properly support Azure OpenAI users.