I recently noticed a $5 charge on my account for usage of models other than GPT-4o-mini. However, I exclusively used GPT-4o-mini for my work, and my total token usage was minimal, around 1000 tokens.
Here are some details about the issue:
Model Used: GPT-4o-mini only.
Token Usage: Approximately 1000 tokens.
Unexpected Charge: $5 for other model usage, which I did not initiate.
Could you please investigate this issue? I would like clarification on why I was charged for models I did not use and how this can be resolved.
You can get more information about usage by going to the “activity” tab, selecting all models in the drop-down that have been used. This should tell you the actual model employed (by tokens, not cost).
If you genuinely have no idea, I would go through user (billing) keys and project keys and revoke and replace them all, along with changing the password of the account access method.
Then you can think about how you may have exposed your API key to adversaries constantly looking for them in insecure code, in committed or shared repositories, etc. Keys should never be placed in code, only be employed by environment variables. Client software used by others should never be making calls direct to OpenAI, only to your API server for servicing chats.