Two weeks. Twenty-plus support emails. Silent downgrades the entire time.
I’m done being polite about this.
OpenAI routes paying accounts to cheaper models without telling you. Your
UI shows the model you picked. The server runs a different one. You pay
the highest tier for it anyway.
When you report it in good faith - because why wouldn’t you, it looks like
a bug - they tell you it’s an internal technical issue and it will be
resolved soon. They know exactly what it is. “Soon” never comes. The bill
comes every month.
I spent two weeks on this. I sent more than twenty emails. What I got back
was templates. What I have left is anger and humiliation.
So here’s what I actually think, and I’m not going to dress it up:
You are not customers to them. You are capacity to be managed. When it
suits them, your model gets swapped and you are never told. When you
complain, you get routed to a queue that exists so that you stop being a
problem in public.
To everyone out there collecting logs, capturing packets, building test
harnesses to prove what’s happening: you look like monkeys in a lab.
Nobody on the other side cares whether you can actually use the product.
You are livestock, and the experiments you run to prove it are free
QA for the company that’s doing it to you.
And you can’t do anything about it. You either keep paying and pretend
you didn’t notice, or you walk away and go back to doing everything by
hand. Those are the options they’ve left you.
I chose to walk.
I cancelled Pro 20x. I’m not coming back. And every time someone mentions
OpenAI, I’m going to tell them exactly what happened to me.
I put up with the contempt and the silence for two weeks. I ran out of
patience before they ran out of templates.
Not goodbye. Never again, OpenAI.