It’s so expensive ! Is this expected?
I think there’s a serious issue with request accounting here.
According to the model section, Opus 4.6 Fast / Thinking is marked as 6×. That implies one user message should cost around 6 requests, not hundreds.
However, as shown in the screenshot, I sent one very small, simple message, just to test the model, and it consumed ~232 requests in a single entry.
To be clear:
-
I sent one message
-
The model UI indicates 6×, not 200×+
-
No batch, loop, agent workflow, or repeated prompts
-
Yet over 200 requests were deducted at once
If internal backend calls are counted separately, this is not clearly communicated and directly contradicts what the model selector implies. From a user’s perspective, this looks misleading and unfair
3 Likes
Hey @Jonas_Mikkel_Mind!
Max mode bills at token-based prices (the total cost of tokens is estimated at an equivalent number of requests) on request based plans. Alongside fast mode, it is just that expensive.
1 Like
you were lucky ^^
look at this ^^
But damn it was fast and fixed 4 errors without asking (introduced 1 though)
1 Like
will take a look, Thank you Artemonim