WARNING: OpenAI Support Has No Way to actually investigate things.

I pay for theChatGPT Pro tier mainly for Codex.

After a usage reset from Tibo, my account was restored to 100% and ChatGPT showed August 7 as my next weekly reset date. I planned my work around that date.

Then, when I was almost out of usage, the reset date suddenly changed to August 12.

So I contacted support with a very simple question:

“Codex showed me Aug 7. Then you showed me Aug 12. What happened?”

After multiple AI replies, and asking for an escalation, I finally got their actual answer:

They cannot see the event-by-event reset history, cannot determine what caused the date to change, cannot escalate it to a team that can check, and cannot restore or compensate the missing usage.

So the final situation is basically:

OpenAI: “Your reset is Aug 7.”

Later:

OpenAI: “Actually, Aug 12.”

Me: “Why?”

OpenAI: “We don't know, we can't find out, and we can't compensate you.”

That is an unbelievably strange level of support for a paid product where usage limits directly affect whether you can actually use the service. I was obviously happy for the free reset, but I cannot keep paying for a product that can just hallucinate a reset date. I switched from Claude where this didn't happen once.

The funniest part is that support explicitly acknowledged that I relied on the reset date shown in ChatGPT when planning my Codex usage. They just have no way to explain why their own system changed it.

These are the last two messages if anybody wants to read through them: https://imgur.com/a/wrVKdwR

EDIT: A few people seem to think I'm complaining about not getting another free reset. I'm not.

I wasn't asking OpenAI for free tokens. I was asking them to check the logs and tell me why the reset date their own app showed me changed.

Their final answer was basically that they can't access the event history, can't determine what caused it, and can't escalate it to anyone who can.

That's the part I find concerning. Bugs happen. What bothers me is a company handling a paid service telling me they can't investigate what happened to my account.

If something more serious ever happens with billing, usage, or account access, it's hard to have much confidence in support when they can't even check the relevant logs.

I'd genuinely rather have stricter but predictable limits with anthropic than more generous limits I can't reliably plan around.