Same thing. After changing keys - it reappears.
Moreover, it went way over my monthly set limit. Charged the credit card, support is not replying. I’m repeating, issues reappears even after changing the API keys, deleting all uses:
Unexpected Charges for “Other Models” - October 25th
My usage report shows a $23.45 charge for “Other models” on October 25th, which is unexpected. How to identify which specific model responsible for this charge and provide usage details? I’d also appreciate any guidance on preventing similar charges in the future.
Hey guyz,
- My issue is partially resolved: I downloaded detailed usage from https://platform.openai.com/organization/usage/activity
and identified that one of my developers was responsible for API calls in question. - However, API limits were surpassed by at least $200 (original limit was $400 and it went over $600 within 12 hours).
So, this is not OpenAI bug - API “other model” usage was our own unfortunate “accident”.
Guyz, thank you for your spport, guidance, time and effort. This means a lor to developers like us.
1 Like
I have deleted all my API keys and set a budget limit, and still the charges continue.
- Mystery usage for me started Oct 14.
- Usage exports has API key column blank
- Usage report for Oct shows unexplained ~3M tokens on various GPT-4 models.
- Usage report for Nov has activity <1k tokens on one single day, but cost of $4-14 each day (again API key is blank)
- First I set all API keys to read-only. Usage costs continued.
- Then I set a budget limit of less than what I’ve already spent. Usage costs continued.
- Yesterday I deleted all API keys; it shows cost today (again no entry on activity report)
Sent message to support yesterday, but no response after 24h.
Today I have canceled my plan and deleted all billing methods; we’ll see if that helps. Next step will be credit card chargeback.
2 Likes
Seems we are not alone:
Check “user” API keys, legacy ones that have no permissions, and before you had to generate a new one to delete existing ones.
https://platform.openai.com/settings/profile/api-keys
Then check team members to see if someone invited themselves to org, and can spend freely.
1 Like
Thanks, I have revoked all those keys now too.
Why isn’t that page linked to from the normal API key or usage pages?
Doesn’t explain the other issues though, like the costs with no usage at all, even untracked, or usage past budget limits.
You have to “view entire organization” usage, and then also on the usage “activity” page, in the drop down, select all the models that appear, to see all their usage. By default only some are checked.
You can export a list from that page to audit what is being called, like just a two day range that is suspect.
This page will give an overview of all keys across projects besides the separate user keys just linked: https://platform.openai.com/settings/organization/api-keys
Thanks to all who posted that they have issues with usage overages.
I too am seeing suspicious usage charges for a web app I recently deployed. It just doesn’t have that much traffic or large input tokens.
Took your advice, created a new key, deleted the old and allowed only one model, gpt-4-turbo, under Project>Limits>Model usage. Will update my usage findings here.
1 Like
Over $800 in charges !
Overnight. Please help
1 Like
We’ve had the same issue since yesterday - sudden surge in API requests to high-cost models (GPT 4/4-turbo etc) bundled under ‘other models’ in the billing which continues to accrue costs beyond the limits we’ve set. We don’t use any of these models in our applications so assuming an API key has been compromised and acting accordingly. @gokulraya please include us in any messaging/investigations if possible!
The same happened to us. On the night of November 10 to 11, there was a sudden and significant consumption by “Other Models” until the hard cost limit was reached, even though no one on the team used the API during that time.
The activity logs show that several old models were allegedly used in parallel, which were then probably summarized as “Other Models” in the costs. However, the token counts listed for each of these do not add up to the extremely high total that was invoiced. Furthermore, nobody on the team actually uses these old models anymore, so nothing should have appeared here in the first place.
And a further analysis of the activities showed that there had already been smaller postings to “Other models”, i.e. old models that no one in the teams uses, in October. It just wasn’t noticed straight away because the amounts were small. So there has probably been a problem for some time.
What I can already say, this is definitely a bug on openai’s part. I hope this will be clarified soon by the support team, whom I already contacted yesterday. If you cannot trust that the billing will be done properly, it becomes difficult to continue working effectively. Please help soon!
1 Like
I was able to stop the over charges by creating a new key, deleted the old and allowed only one model, gpt-4-turbo, under Project>Limits>Model usage.
Were you able to resolve the over charges?
Yesterday OpenAI has refunded all the overcharges, though they did not really explain the issue–just got some copy/paste on how to view usage and API keys.
2 Likes
Dear OpenAI Team,
We are being charged for other models, even though we have blocked access and are only using GPT-4o mini and audio models. Despite this, we were unexpectedly charged $20.
I am unsure why this is happening since the models are supposed to be restricted.
Kindly look into this issue and process a refund for the amount charged.
Thank you for your prompt attention to this matter.
1 Like
I just had a heart attack seeing a massive charge for something I never used! over $20 in a single day? I never spend that much in a month!
Any updates regarding this issue? That happened to me again in December.
In November, I experienced a serious incident with my OpenAI API usage:
- Models we never used were billed. We checked thoroughly for any credential leaks and found none, yet all AI models— including ones like GPT-3.5 (which we have no reason to test right now) and image generation (which we never use)—showed activity. This is very strange.
- We exceeded our monthly limit by a huge margin. Despite having a $200 spending limit set (and receiving an alert email just before hitting that limit), we ended up being charged far beyond this amount.
I noticed the issue in late November and immediately deleted all API keys. However, by the end of December, over $8,300 had already been charged to our credit card. This is a severe financial blow, and it will have an impact on our company’s financial statements if not resolved promptly.
Since late November, I have reached out to Support multiple times. They initially said they would look into it, but then communication stopped. Even after opening new tickets, I have heard nothing back.
- I need an explanation for why the spending limit was exceeded before I can safely use the service again. However, more urgently, I want a refund for any charges beyond the $200 monthly cap. Over the past six months, our monthly usage always stayed within $200. November’s charges are clearly abnormal.
I am posting this here because I’m unsure if the official team will see it—but I’m hoping it will catch someone else’s attention. I truly need help resolving this issue.
1 Like