AzureAI
Guide to deploying and using the TimeGEN-1 model as an Azure AI endpoint.
Azure Deployment Note
The foundational models for time series developed by Nixtla can be deployed directly to Azure subscriptions. This guide explains how to quickly start using TimeGEN-1 as an Azure AI endpoint. If you currently use the nixtla
library, the Azure deployment works as a drop-in replacement—simply adjust the client parameters (endpoint URL, API key, and model name).
Deploying TimeGEN-1
Time Gen on the Azure Portal
Using the Model
After you have successfully deployed TimeGEN-1 and ensured you have permission to access its endpoint, you can interact with the model as you would with a standard Nixtla endpoint.
Ensure you have your deployment URL and API key ready before proceeding.
How to Use
Install the Nixtla Client
This installs the official Nixtla Python client library so you can make forecast requests to your Azure AI endpoint.
Set Up Your Environment
Make sure you have the following environment variables properly configured:
-
AZURE_AI_NIXTLA_BASE_URL
-
AZURE_AI_NIXTLA_API_KEY
Initialize the Nixtla Client
Here, we create a new client instance using your Azure endpoint URL and API key.
Make a Forecast Request
Replace the ellipsis (…) with your specific forecasting parameters and then call the endpoint to get predictions.
Key Concept: Drop-In Replacement
Because TimeGEN-1 on Azure uses the same API structure as the Nixtla library, you only need to switch out the base URL, API key, and model name. Your workflow remains unchanged.
Key Concept: Seamless Integration
Deploying TimeGEN-1 to Azure allows you to leverage Azure’s scalability, security, and management tools directly for your time series forecasting needs without altering core application logic.
Tip: Remember that you can use any Azure-supported authentication or security measures to further protect your endpoint, such as Azure Key Vault for managing secrets or role-based access control for restricting usage.