I noticed a significant increase in my Hugging Face Inference API costs. Last month, I made 569 requests, which cost $0.54. This month, with 690 requests, the cost jumped to $6.38—a 10x increase per request.
Previously, each image inference cost around $0.005, but since March 24, it has increased to $0.05 per image.
I’d like to understand:
What could be causing this sudden price jump?
How can I check which GPU instance is being used?
Is there a way to optimize costs or choose a lower-cost option?
I noticed the “Pay as You Go” (PAYG) update was announced 14 days ago, but my cost increase started on March 24. Last month, 569 requests cost me $0.54, but this month, 690 requests already cost $6.38—a 10x increase per request.
Previously, each image inference cost around $0.005, but since March 24, it has jumped to $0.05 per image.
Could this be related to the PAYG change?
Has anyone else experienced the same issue?
Did anyone find a solution?
I’ve already emailed Hugging Face support, but I’m curious if others are facing this too. Any insights would be appreciated!