Severe ChatGPT Pro Billing Bug: Support sends irrelevant Codex templates to cover up double-charging (Case: 08252537)

Hi everyone,

I am writing this post in the hope of catching the attention of an OpenAI Community Manager or Tier 2/3 Billing Support. For the past two months, I have been caught in a frustrating loop with frontline customer support regarding a clear, system-level billing bug, and the experience has been incredibly disappointing.

The Core Issue: Backend Double-Charging

In early January 2026, while using add-on prepaid credits for the ChatGPT Pro model in the web interface, I noticed my credits disappearing twice as fast as expected.

Upon exporting my official usage logs (CSV) directly from the OpenAI Dashboard, I found a definitive backend concurrency/duplication bug:

• According to OpenAI’s official Rate Card, GPT-Pro costs a flat rate of 50 credits per 1 message.
• However, my logs show multiple entries where a single prompt-response interaction was recorded with a usage_quantity of even numbers (2, 4, 6 counts) at the exact same timestamp, resulting in 100, 200, or 300 credits deducted at once.

The Support Failure: Complete Gaslighting

I have been in communication with support for two months (referencing Case Numbers: 08252537 , 08043456 , and 06096891). I provided the official exported CSV, receipts of my $20 credit purchases, and current screenshots.

Despite the frontline support agents explicitly acknowledging in writing that the rate card is a flat 50 credits/message, they refuse to address the duplication in their own logs. Instead, they have resorted to:

1. Asking me for a “screen recording” of a bug that happened months ago on an older model version (GPT-5.2), which is technically impossible to reproduce now that my balance is 0 and the platform has migrated to GPT-5.5.
2. Asking a standard web UI user for API backend “Request IDs”.
3. (Most absurdly) Sending me copy-pasted Codex troubleshooting templates, telling me to “reduce the size of my AGENTS.md” and “limit MCP servers” to make my limits last longer. My issue is a flat-rate message billing duplication error , not an API/Codex context token optimization problem!

Request

I am not asking for a free handout. I am demanding that OpenAI investigate its server logs, acknowledge this database duplication bug, and restore the prepaid credits that were wrongfully deducted from my account.

If any OpenAI staff member here can escalate this ticket to a technical billing supervisor who actually understands the difference between ChatGPT web billing and Codex API tokens, I would be deeply grateful.

Thank you.