1.We understand that models are hosted in confidential infrastructure not disclosed in public domain. Is every model instance or deployment hosted in a separate dedicated infrastructure or is there a shared infrastructure within Microsoft that commonly hosts all instances?
2.When building applications based on Azure OpenAI, how do we guarantee performance/responsiveness for varying loads/usage patterns, if we do not have control over selection of the hosting hardware?
Thanks in advance.
Please share the answer related to the question.
