Greetings
A few weeks ago I got an email saying that fine tunning was going to be free till September X, so I decided to try it out and I did. After that while monitoring my cost (because i’m trying to get a feel of how much this app will cost in the real world) I noticed that it charged me some cents for fine tunning despite it being free, I paid it no mind.
Fast forward, i’m using the fine tuned model in my desktop app, so it’s getting no real usage apart from me testing various things. I did the fine tunning on the 20th of September, havent touched it again, yet i’m being charged for “fine tunning models” as recent as the 30th.
Do you get charged for using the fine tuned model? I’m being charged for both gpt4(I expect that) and for fine tunning which I havent done since the 20th of August. Please assist me, am I the one understanding this pricing wrong?
Please refer to the two images I posted. One shows I did fine tunning on the 20th of August, but my usage costing is still billing me till the 30th for fine tunning.
Yes, you do get charged for using the fine-tuned model as per the official pricing here. Only the fine-tuning itself is free up to the defined thresholds of 2M and 1M training tokens for gpt-4o-mini and gpt-4o, respectively. When you calculate the training tokens, the following formulate applies:
(base training cost per 1M input tokens ÷ 1M) × number of tokens in the input file × number of epochs trained
Let me clarify, I am aware I get charged for using the model. but 1) I was charged for fine tunning a free model (gpt-4o-mini) and I was not past the token limit.
Secondly, I fine tuned on the 20th. Yet today is the 30th and every day I keep getting charged for fine tunning and i’m doing no such thing.
Look at my first image and my latest image. I did two sucessful finetunnings on the 20th.
Why is there a charge on the 30th? for Fine tunning?
You can check the activity tab that provides details on the token consumption:
There you should see two sections relating to fine-tuned models, one relating to the use of your individual fine-tuned models and one relating to fine-tuning process itself. The latter looks as follows:
You should check there if it shows trained tokens for August 30th.
In the upper part of the activity tab you can also inspect the token usage of your individual fine-tuned models. You can validate there whether there are any charges for August 30th.
Ok so from that page, it just shows the two jobs and the token correct on the 20th. Nevertheless, money is still being deducted from me, and I dont know from where.
I’m afraid your only option is to submit a detailed message via the Support Chat. They will then get back to. Given the high volume of requests, this may take some time.
It looks like being charged for the use of your fine-tuned model. That’s perfectly normal. Only the training is free up to the defined token limits per day. Whenever you send an actual message to your fine-tuned model, you incur charges.
To me it looks like there is no issue here unless I am misinterpreting the information you provided.
The price of a fine tuned model is only slightly higher than one that’s not. Out of 10 “suggest me X” type questions. I find it hard to believe I’ve generated 10 cents worth of charges. Secondly, if that’s the case, why is GPT-4o mini model still showing?
