Banner & AI API Preferences

Azure OpenAI API Key

Learn how to Fetch your Azure OpenAI API Key

Microsoft Azure Government is an essential addition to Innoslate's cloud infrastructure, providing a highly secure and compliant platform specifically designed to meet the unique demands of U.S. government agencies. Operating from advanced, fortified data centers in regions such as US Gov Virginia, US Gov Texas, and US Gov Arizona, Azure Government delivers robust security measures and adherence to stringent compliance standards, making it an ideal solution for managing sensitive and classified data. Significant milestones such as the recent approval of the Azure OpenAI Service under the FedRAMP High Authorization for Azure Government, which empowers federal agencies to tap into cutting-edge AI technologies while satisfying rigorous security protocols. Additionally, Azure OpenAI has secured authorization for use within the Department of Defense (DoD) at Impact Level 4 (IL4), Impact Level 5 (IL5), and Impact Level 6 (IL6) Provisional Authorizations.
With the integration of GPT-4o into the Azure OpenAI Service, now available within both FedRAMP High and DoD IL4, IL5 & IL6 frameworks, government agencies can leverage its powerful multimodal capabilities—spanning text, vision, and audio processing. This opens the door to transformative applications, such as enhanced natural language processing, improved decision-making, and increased productivity, all while maintaining the highest levels of data security and regulatory compliance required for government use.

Signing Up and Providing Your Information

The journey begins with creating or logging into your Microsoft account. If you don’t already have an Azure Account, you’ll need to sign up by providing your personal details. This step is crucial for establishing your identity and ensuring secure access to Azure services.

After logging in or creating a new account, you’ll be prompted to enter your information, including your country/region, full name, email address, phone number, and physical address. 

Azure_CreateAccount

Providing a Payment Method

To access Azure services, including the Azure OpenAI API, you’ll need to provide a valid credit or debit card. Azure offers an incentive for new users: a $200 credit to use within your first 30 days. This credit allows you to explore various services without immediate charges. After 30 days, users will be moved into a pay-as-you-go or batch account. During this step, you’ll see a form requesting your cardholder name, card number, expiration date, and CVV.

Azure_Payment

Exploring the Azure Landing Page

Once your account is set up and payment method is verified, you’ll land on the Azure dashboard.  The landing page features sections like Azure services (including AI, virtual machines, storage, and more), resources (where you can view recently accessed or favorited items), navigation (to manage subscriptions, resource groups, and more), and tools (such as Microsoft Learn, Azure Monitor, and Cost Management).

Azure_LandingPage

Before proceeding, please ensure you have an active subscription available. An active subscription is necessary to access and utilize Azure OpenAI services. If you’re unsure, check your subscription status in the step 4.

Managing Your Subscriptions

Next, navigate to the “Subscriptions” section from the Azure portal’s navigation menu. Here, you’ll see a list of all available subscriptions associated with your account, along with their details. Each subscription includes information such as the subscription name, ID, your role (e.g., Owner), current cost, secure score, parent management group, and status (e.g., Active).

Azure_Subscriptions-1

If you don’t see an active subscription or need to create one, use the “Add” option to set up a new subscription. Ensure your subscription is active to proceed with accessing Azure AI services.

Accessing Azure AI Services

From the Azure landing page, locate and click on “Azure AI services” in the services menu. This section houses a collection of AI-powered tools and capabilities designed to enhance your applications. Azure AI services include options for natural language processing, computer vision, and more, setting the stage for integrating advanced AI into your projects.

Azure_AIServices

Setting Up Your Azure OpenAI Account

Within the Azure AI services section, navigate to the “Azure OpenAI Account” option. This is where you’ll begin the process of creating and configuring your Azure OpenAI account, which is essential for accessing the OpenAI API and its powerful language models. The Azure OpenAI Account allows you to perform a wide variety of natural language tasks, such as text generation, translation, and more, making it a versatile tool for developers and businesses alike.

Azure_OpenAiAccount

Creating Your Azure OpenAI Instance

After navigating to the “Azure OpenAI Account” section, you’ll need to create an Azure OpenAI instance to access its services. This step involves configuring several key details to set up your instance correctly:

  • Subscription: Select your billing subscription from the dropdown menu. For example, choose “Azure subscription 1” if it’s your active subscription. This ensures that any usage costs are billed to the correct account.
  • Resource Group: You can either create a new resource group or reuse an existing one. Resource groups help organize and manage your Azure resources. For instance, you might name a new group “SPEC_Demo” to keep your OpenAI resources organized.
  • Region: Choose a Region where your Azure OpenAI instance will be hosted, such as “East US 2.” The region is significant because access to certain AI models and features can vary by location, so select one that best suits your needs and availability of services.
  • Name: Provide a unique name for your instance, like “spec4,” to identify it within your Azure environment.
  • Pricing Tier: Select a pricing tier that aligns with your usage requirements. For example, you might choose “Standard S0” for a balanced option suitable for most applications. You can view Full Pricing Details to ensure it fits your budget and needs.

Azure_CreateAIInstance

Once these details are configured, proceed to the next steps to finalize your instance setup. 

Configuring Network Security

In the network configuration step, you’ll determine how your Azure OpenAI instance can be accessed over the network, ensuring security and control over your AI resources. The available choices are:

  • All networks: Including the internet, can access this resource. This option allows unrestricted access from any network, which is suitable if you need broad accessibility.
  • Select networks: This option lets you limit access to specific networks, providing more control and security for your instance.
  • Disabled: No networks can access this resource. This option restricts all network access, requiring private endpoints for secure, exclusive access.

Azure_NetworksSelect the option that best matches your security and accessibility requirements, then move forward.

Adding Tags to Your Resource

Tagging is a useful way to organize and categorize your Azure resources, making management and billing easier. In this step, you’ll create name/value pairs to label your Azure OpenAI instance.

Azure_Tags

  • The tags interface allows you to enter a name (e.g., “Environment”) and a corresponding value (e.g., “Production” or “Development”). These tags help you categorize resources and view consolidated billing by applying the same tags across multiple resources and resource groups.

Note: If you create tags and later change resource settings on other tabs, your tags will automatically update to reflect those changes.

    Reviewing and Submitting Your Instance

    Before finalizing the creation of your Azure OpenAI instance, you’ll review all the details you’ve configured and submit your setup for processing.

    Navigate to the “Review + submit” section, where you’ll see a summary of your configuration, including:

    • Basics: Subscription, resource group, region, name, and pricing tier (e.g., Azure subscription 1, SPEC Demo, East US 2, spec4, Standard S0).
    • Network: Your chosen network access type (e.g., “All networks, including the internet, can access this resource”).
    • Tags: Any name/value pairs you’ve added (e.g., Environment: Production).

    Azure_Review

    Once you’ve reviewed everything, click the “Create” button to submit your configuration and initiate the creation of your Azure OpenAI instance.

    Navigating to Azure OpenAI Resource

    After creating your Azure OpenAI instance, return to the Azure portal’s main dashboard and navigate to the “Resources” section. Here, you’ll find a list of recently accessed or favorited resources. Locate and select the “spec4” Azure OpenAI resource, which you created earlier. This resource represents your Azure OpenAI instance, and selecting it will take you to its dedicated management page.

    Azure_NavigateResource

    Resource Details & Preparing for API Deployment

    Once you’ve opened the “spec4” resource, you’ll see a detailed overview of its configuration. This page includes essential information such as the resource group (SPEC Demo), status (Active), location (East US 2), subscription (Azure subscription 1), and more.

    • Key Actions (Highlighted in Red): Pay special attention to the “Endpoints” and “Manage keys” options, both of which are critical for setting up your API access:
      • Endpoints: Click here to view the endpoints associated with your Azure OpenAI instance. Endpoints are the URLs you’ll use to send API requests to your deployed models.
      • Manage keys: Click here to access and manage the API keys for your instance. These keys are necessary for authenticating your API requests and ensuring secure access to the OpenAI models.

    Azure_ResourceDetails-1

    • Navigation to Azure AI Foundry (Highlighted in Blue): Look for the “Go to Azure AI Foundry portal” button. This is a pivotal step, as the Azure AI Foundry portal is where you’ll configure and deploy your AI models, setting the foundation for using the Azure OpenAI API effectively.

    Click the “Go to Azure AI Foundry portal” button to proceed to the next phase of your setup.

    “Deployments” Tab in Azure AI Foundry

    Upon entering the Azure AI Foundry portal for your “spec4” resource, navigate to the “Deployments” tab, as highlighted in the red box. This tab is where you’ll manage and deploy AI models for your Azure OpenAI instance. If no deployments exist yet, you’ll see a message indicating “No deployments to display,” prompting you to create a new deployment.

    Azure_FoundryDeployments

    Selecting the “Deploy Model” Option

    In the “Deployments” tab, locate and click the “Deploy model” dropdown menu (highlighted in the image). This dropdown provides options to deploy different types of models, such as base models, fine-tuned models, or models from Azure ML. Select “Deploy base model” to begin the process of deploying a pre-trained model, which is a common starting point for most users.

    Azure_DeployModel

    Selecting a Model for Deployment

    After choosing “Deploy base model,” a popup titled “Select a model” will appear. This interface allows you to choose the specific AI model you want to deploy for your application. Here’s how to proceed:

    • The popup displays a list of available models, such as “gpt-40-mini” and others (e.g., gpt-40, gpt-40-audio-preview, etc.). Each model is categorized by its inference tasks, like chat completion, audio generation, or real-time tasks.
    • Select “gpt-40-mini” by checking the box next to it. This model is optimized for low cost and latency, making it ideal for tasks like chat completions, real-time text responses (e.g., customer support chatbots), and applications that chain or parallelize multiple API calls.
    • You can toggle the “Show description” option to view detailed information about each model, such as its capabilities and use cases. 

    Azure_SelectaModel

    Once you’ve selected your model, click “Confirm” to proceed with the deployment configuration.

    Reviewing Model Deployment Details

    After confirming your model selection, you’ll see a popup with detailed deployment settings for the “gpt-40-mini” model. Here’s a breakdown of all elements in this popup:

    • Deployment Name: This field requires you to enter a unique name for your deployment. By default, it may suggest “gpt-40-mini,” but you can customize it to something meaningful, like “chatbot-deployment,” to identify it easily.
    • Deployment Type: Choose the deployment type that suits your needs. The default option is “Global Standard,” which offers pay-per-API call pricing with the highest rate limits. This type processes data globally outside of your resource’s Azure geography, but data storage remains in the AI resource’s Azure geography (e.g., East US 2). You can learn more about deployment types by clicking here.
    • Deployment Details:
      • Model Version: Displays the version of the model you’re deploying, such as “2024-07-18” for “gpt-40-mini.” This ensures you’re using the latest or specified version of the model.
      • Capacity: Indicates the model’s capacity, measured in tokens per minute (TPM). For “gpt-40-mini,” this might be set to 30K tokens per minute, defining how many tokens the model can process in a minute. Quota increase forms available upon higher demand for TPM.
      • AI Resource: Shows the associated Azure OpenAI resource, in this case, “spec4.”
      • Resource Location: Confirms the Region of your resource, such as “East US 2,” ensuring data residency compliance.
      • Content Safety: Specifies the content safety settings, such as “DefaultV2,” which helps mitigate harmful use of the model by filtering or flagging inappropriate content.
    • Actions: You have options to “Customize” the deployment settings for advanced configurations or “Deploy” to finalize the deployment. The “Cancel” button allows you to exit without creating the deployment.

    Azure_FinalizeDeployment

    Review all details carefully, make any necessary adjustments, and click “Deploy” to create your model deployment. This step is the most critical part of enabling AI on your Innoslate platform.

    Reviewing Deployed Model Details

    After deploying the “gpt-40-mini” model in the Azure AI Foundry portal, you can view its details by navigating to the model’s deployment page. This page provides a comprehensive overview of your deployed model, ensuring you have all the necessary information to integrate it into your applications.

    • Target URI: This is the URL you’ll use to send API requests to your deployed model. This endpoint is critical for connecting Innoslate to the model.
    • Key: This is the API key associated with your deployment, displayed as a series of asterisks for security (e.g., ************************). You can reveal the full key by clicking the eye icon, but ensure you handle it securely as it authenticates your API requests.
    • Model Name: The precise model name, “gpt-40-mini,” is displayed under “Model name.” This name is essential for configuring your application to use the correct model and must match exactly when setting up your API preferences.

    Azure_DeploymentOverview-1

    Additional details on this page include:

    • Deployment Info: Shows the deployment name (“gpt-40-mini”), provisioning state (“Succeeded”), creation and modification dates, and the user who created and modified it.
    • Rate Limits: Indicates the model’s capacity, such as 30,000 tokens per minute (TPM) and 300 requests per minute, helping you manage usage and performance.
    • Model Version and Lifecycle: Displays the model version (e.g., “2024-07-18”), lifecycle status (“GenerallyAvailable”), creation date, update date, and model retirement date (e.g., July 19, 2025).
    • Monitoring & Safety: Includes the content filter setting (e.g., “DefaultV2”), which helps mitigate harmful use of the model.
    • Useful Links for Application Development: Provides resources like a code sample repository and tutorial links to assist with integration.

    Take note of the Target UR, API key and Model Name, as you’ll need these to configure your API preferences in the next step. You can also use the “Open in playground” button to test the model interactively before integration.

    Configuring [Organization]’s Preferences for Azure OpenAI

    To finalize your Azure OpenAI API setup, you’ll need to configure your organization’s preferences, typically accessed through an admin interface like “SPEC Innovations’ Preferences.” This popup appears when an administrator is setting up AI API preferences for the Azure OpenAI provider. Here’s a detailed breakdown of each section in the popup:

      • Provider: A dropdown menu where you select the AI provider. In this case, “Azure OpenAI” is selected, confirming that Azure OpenAI is the chosen service for your organization’s AI needs.
      • Base URL: This field requires the base URL for your Azure OpenAI instance. Enter the Target URI from Step 17 to establish the connection to your deployed model.
      • API Key: Input the API key from Step 17 (e.g., the masked key you revealed earlier) into this field. This key authenticates your requests to the Azure OpenAI API, ensuring secure access. The eye icon allows you to toggle visibility of the key for verification.
      • Embedding URL (Helpbot): This field specifies the URL for embedding models used by the Innoslate Helpbot. Enter text-embedding-3-small URL from Azure Foundry deployment, which points to a specific embedding model for generating vector representations of text.
      • Chat Model: This field requires the exact name of the chat model you deployed. Input “gpt-40-mini” (highlighted in blue from Step 17) to ensure your organization uses this model for chat completions. This must match the model name precisely for compatibility.
      • Token Limit: This field sets a limit on the number of tokens your organization can use. In the image, it’s set to “0,” indicating no limit or a default setting that needs adjustment based on your usage policy.

    Azure_InnoAPI

    Once all fields are correctly filled, save your preferences to apply the Azure OpenAI configuration to your organization’s AI systems.

    You’ve successfully completed the process to set up and configure your Azure OpenAI API Key. By following these steps—starting from signing up for Azure, creating your OpenAI instance, deploying a model like “gpt-40-mini,” and configuring your organization’s preferences—you’re now ready to integrate powerful AI capabilities into Innoslate.

     

    To continue learning about Organization Preferences, Click Here.

    (Next Article: AskSage API Key)