Deploying LLM model in Azure AI Studio in 5 minutes

This article shows how to deploy a ChatGPT-4o in Azure AI studio in ~5 minutes time.

Prerequisites:

1. Azure account ( 5 min onboarding ).
2. Active Azure subscription like this one: (Microsoft 4 minute learn video)

Azure subscription is a way of Azure to connect the resources to the billing and payment methods.

Here are the two steps to deploy the gpt-4o:

Step 1: Create Azure resource

It has 7 simple sub steps :

a. Browse to https://ai.azure.com/

b. Select “Resources and keys”

c. Press “Create new OpenAI resources”

d. Define the resource details according to this step-by-step wizard below:

—Provide the your Azure “subscription” name (which is connected to your billing and payment).

— Select the “resource group”, if not having any, you can create new at “Create new” button in this page.

— Set your resource name, which will be used later for model deployment, therefore give it a meaningful name say ai-resource-your-username.

— Select the pricing tier.

e. Define the network of the resource:

f. You can skip the tags.

g. Review the resource in the last stage of the wizard, and push “Submit”.

Now you have an available Azure resource which we can be used later for LLM deployment.

Step 2: Deploy the LLM using Azure resource

a. Browse to https://ai.azure.com/

b. Go to “Model Catalog” and select gpt-4o “Chat completion” model

After selecting the model you will see the screen blow.

Press “Deploy”

c. You will be asked for the Azure resource (which we prepared in #Step 1), select the resource and push “Deploy”

d. Now you can see your deployed model in the “Resources and Keys” section at https://ai.azure.com/.

Now the gpt-4o is deployed at the Endpoint shown in the screen above (marked in purple). You can find the AI API key at the same screen to the right (marked in purple).

Conclusion

Azure AI Studio has most of the recently released LLMs. In order to explore their capabilities you need to run the process described above.Once the LLM deployed, you can showcase the LLM in the chat Playground or use it in the your applications by defining the following API details in your app:

      - GLOBAL_LLM_SERVICE=AzureOpenAI
      - AZURE_OPENAI_ENDPOINT=https://your-endpoint-name
      - AZURE_OPENAI_CHAT_DEPLOYMENT_NAME=gpt-4o
      - AZURE_OPENAI_API_KEY=your-key