Hierarchy and relationships between Azure OpenAI, Azure Hub, Azure AI, Azure AI Project, Azure AI Services...

Question

I surely can't be the only person thrown by the spaghetti of different AI services provided by Azure? Where do I start? I set up OpenAI months ago. It worked well. Now I want to add managed Llama 3, Mixtral, etc., services but have become lost in the below:

1. Azure OpenAI
2. Azure Hub
3. Azure AI
4. Azure AI Project
5. Azure AI Service

Thanks!

Answer 1

@Girish Gupta If you are using Azure OpenAI resource that was previously setup you can continue to use the same using Azure OpenAI studio.

If you want to try out the new Azure AI studio which integrates all Azure AI & ML services you will use the hub & project analogy where hub is a top-level resource in AI Studio where resource provider is Azure ML workspace of kind hub.

AI project is a child resource of the AI hub where the resource provider is Azure ML workspace of kind project. If you create multiple projects under the hub you will be creating multiple Azure ML workspaces which share the security same settings as hub and they can share artifacts in workspaces like data connections.

Azure OpenAI and Azure AI services are connections to the project where you can basically use the endpoints of these individual services and use them in your projects.

You can refer the architecture reference from AI studio for more details.

If you have to deploy a Mistral model for example, you will be using your project and deploying the model in that project workspace or the Azure ML workspace that was created when you created the project.

Other option to create a mistral or llama model without having to use the AI studio or the above architecture is to use your existing Azure ML workspace from ML studio, navigate to Model catalog page and select the model that needs to be deployed and deploy it as managed endpoint.

If you do not have an existing Azure ML workspace, I would recommend trying AI studio and the new architecture as this will integrate all your future Azure AI service needs under one portal for easier integration. Thanks!!