I am planning to build an application using Bedrock's image generation models: a) Stability Diffusion SDXL Beta 0.8 and b) Titan G1
I want to use both these models on an on-demand basis and I am aware that I will be charged per request. Is there any other hourly or monthly subscription charge if I use the Stability Diffusion SDXL Beta 0.8 foundation model or the Titan G1 preview model?
I have seen some publicly available pricing links on the AWS website and it seems that apart from the per-request pricing, there is an additional real-time inference and infrastructure charge for Sagemaker. I raised a ticket with AWS Support but their response was not clear.
Can someone please help me understand the pricing estimate with an example for both SDXL Beta 0.8 and Titan G1 models?
Example: What would be total cost payable to AWS if a web application makes 1000 requests for image generation in 1 day (24 hours) for 1024x1024 images (70 steps)? Kindly provide the estimate for both SDXL and Titan models