Skip to main content

Azure AI Foundry

Azure AI Foundry is Microsoft's unified platform for building, deploying, and managing AI models. It provides access to a wide catalog of models from various providers — including OpenAI, Anthropic, Mistral, DeepSeek, Meta, and more — all deployable as managed endpoints within your Azure subscription.

Getting started with Azure AI Foundry​

Prerequisites​

  • An active Azure subscription.
  • An Azure AI Foundry project. If you don't have one, create a project from the Azure AI Foundry portal.

Deploying a model​

  1. Navigate to the Azure AI Foundry portal and open your project.

  2. Click on the Build tab in the top navigation. From the left sidebar, click on Models. You will see the Deployments tab listing your existing model deployments.

    Azure AI Foundry Deployments

  3. To deploy a new model, switch to the model catalog by clicking Discover in the top navigation. Browse or search for the model you want to deploy.

    Azure AI Foundry Model Catalog

  4. Select the model you want to deploy (e.g., claude-sonnet-4-6). On the model details page, click the Deploy button.

    Azure AI Foundry Model Details

  5. A dropdown will appear with deployment options. Choose the appropriate deployment type (e.g., Default settings, Custom settings, etc.).

    Azure AI Foundry Deploy Dropdown

  6. Review the terms of use and pricing information in the deployment dialog. Select your preferred options and click Agree and proceed.

    Azure AI Foundry Deploy Dialog

  7. The model will begin deploying. Wait for the provisioning state to change from Creating to Succeeded.

    Azure AI Foundry Deploying Model

    Azure AI Foundry Deployment Succeeded

  8. Once the deployment succeeds, navigate to the deployment Details tab. Note down the Target URI and the Key — you will need these to configure the LLM in Unstract.

    Azure AI Foundry Deployment Details

Endpoint URL

The Target URI displayed on the deployment details page is the full endpoint URL (e.g., https://<resource-name>.services.ai.azure.com/anthropic/v1/messages). For the Unstract configuration, you only need the base URL up to and including the resource domain — for example: https://<resource-name>.services.ai.azure.com/.

Alternatively, you can use the inference endpoint format: https://<resource-name>-<deployment-name>-serverless.<region>.inference.ai.azure.com/ or https://<region>.inference.ai.azure.com/.

Setting up Azure AI Foundry LLM in Unstract​

Now that we have the endpoint URL and API key from Azure AI Foundry, we can use them to set up an LLM profile on the Unstract platform. For this:

  • Sign in to the Unstract Platform.

  • From the side navigation menu, choose Settings 🞂 LLMs.

  • Click on the New LLM Profile button.

  • From the list of LLMs, choose Azure AI Foundry. You should see a dialog box where you enter details.

    Azure AI Foundry LLM Setup in Unstract

  • For Name, enter a unique name for this adapter instance. Example: azure-ai-foundry-1.

  • In the API Key field, paste the key copied from the Azure AI Foundry deployment details page (refer step 8).

  • For Endpoint URL, enter the Azure AI Foundry endpoint URL. Example: https://unstract-dev-foundry-resource.services.ai.azure.com/. Refer to the info box above for URL format details.

  • In the Model field, enter the model name deployed in Azure AI Foundry. Examples: claude-sonnet-4-6, command-r-plus, mistral-large-latest, gpt-4o.

  • Leave Maximum Output Tokens empty to use the model default, or set a value to limit LLM reply length.

  • Leave Max Retries and Timeout fields to their default values.

  • Click on Test Connection and ensure it succeeds. You can finally click on Submit and that should create a new LLM Profile for use in your Unstract projects.