Generative AI pricing model
| GENCOST02: How do you select a cost-effective pricing model (for example, provisioned, on-demand, hosted, or batch)? |
|---|
Foundation model hosting and inference can be conducted in a variety of ways. Some workloads demand immediate responses, while some can be done in batch. Some are hosted on unmanaged infrastructure, and some are hosted using serverless technologies. The inference and hosting paradigm selected influences total cost and should be done with cost in mind.
Best practices
