Hi,
Since yesterday, I have had a sudden increase in charges for the Gemini API. When I check the billing reports, it shows “Generate content cached content storage token hours gemini 2.5 pro” as the SKU resulting for those charges. I have disabled the API, yet the charges are increasing. How can I stop this?
Thanks!
Hi @Doug_Montgomery, @Doug_Montgomery
Welcome to the Forum,
Please contact the cloud billing support team as they would be the best people to help you. Get Cloud Billing support | Support Documentation | Google Cloud.
Thank you!
Hi @ahmedac @Doug_Montgomery ,
While the Cloud Billing Support team is the right channel for investigating the charges, providing specific technical details helps them to resolve the issue faster.
Based on the information you shared, the SKU "Generate content cached content storage token hours gemini 2.5 pro " points to charges incurred for the storage of cached content.
It’s possible that a cache was created with a long TTL or a distant expire_time. This would lead to ongoing storage charges even if you have since disabled the API.
To help us investigate and resolve this, could you please check the following?
expire_time on each cache is in line with what you expect.For reference, you can also review the context caching pricing details here: Gemini API Caching Pricing.
Thank you!
I have no caches active. In fact, here’s how messed up the billing situation is…
2 screenshots… (actually 3)
One from yesterday when I was talking with the billing support team
and another from today…
Yesterday it was showing billing and charges for the 8th
and today that billing and charges is all of a sudden gone.
Also notice the forecast from the report showing September 1st-8th… That forecast is $11.48 when the actual bill is showing $963.18 for the first week.
Then look at the screenshot from this morning showing September 1st-9th. It shows my bill as $715 and the usage has completely changed on Sep 5th. It is only showing cached usage.
And now this would be funny, but it isn’t.
I just took another look to see if maybe I had some filter or something happening because that last screenshot is only showing cache and not usage.
Here’s the screenshot I just took a moment ago… in the evening of September 9th.
Someone tell me that this is normal and something I can trust?
I think someone needs to refund the extra money they took from my account when they charged me $100 for about $30 worth of usage before I caught this mess.
How’s this one for you?
This morning I’m down to REAL costs… $13.19.
Now, how do I get the refund for the over charges?
I sent an API request with the same distribution, and while I received a normal billing on September 1st, I was charged an absurd rate for SKU 01AD96-3C48C7-152263 (Generate content, cached content storage token hours, gemini 2.5 pro) on September 14th and 15th.
I’ve requested a correction from the billing team, but my billing hasn’t changed. Could you please let me know how things went afterward?
Hey, did you get anywhere with it? We are having the same issue
Same here, 1334% increase over 3rd to 5th September. But for us it was “Generate content input token count Gemini 2.5 Pro short input text” going up without apparent reason.
Same issue here, for the period 16 through 18 September, The costs are for “Generate content cached content storage token hours gemini 2.5 pro” - and our apparent usage is “675,991,468.96 hour”. Last month it was only £98 for the entire month, this is a 9938% increase over that for only three days
Same issue here too. Noticed a spike, then a drop, and today again a massive spike. It’s all related to Generate content cached content storage token hours gemini 2.5 pro on mine.
We have same issue. Billing support told that “everything alright”, although it is 20x more of what we used. Did anyone get refund somehow?
Hi, we have zero cached content and it keeps charging us 500 EUR per day. TTL seems not working for billing. Billing team is redirecting us to this forum.
Can you please help something, looks like critical bug!
We have the same problem! The charges for Generate content cached content storage token hours gemini 2.5 pro have grown tenfold.
We haven’t had any active caches for 3 days already (the cachedContents method always returns an empty list), but the charges are increasing daily and the number of tokens/hour continues to grow. How do we stop this?
I have a specific function that detects and deletes possible orphan caches that I run several times through the lifespan of my code to ensure I do not leave any caches behind, not to mention a very short TTL. This is definitely a bug. I have contacted billing support and they have sent it up the chain. But there are several others experiencing the same thing. We can only hope that the Google team resolves this.
My original three days of huge unexpected charges have now been removed from my usage costs, no word on refunding the payments taken. And now the next three days have the same issue. The bug is clearly ongoing.
Absolutely it is, every time I check my console it’s gone up some more. I deleted my API keys hours ago. This is clearly a bug on their end. I can’t imagine the havoc this is causing on larger projects. No word from anyone about it.
