Deploy Upstage products β Information Extract, Document Parse, and Studio β directly within your Azure environment using managed applications. This deployment method creates dedicated endpoints that run exclusively in your Azure subscription and private virtual network, providing enterprise-grade security and control.
This page covers the shared prerequisites first, then a per-product deployment guide. In the Deployment section below, expand a product β Information Extract, Document Parse, or Studio β to see its deployment guide.
Key Benefits
- Enhanced Security and Privacy. Your data remains within your own Azure infrastructure, ensuring complete control over data flow and eliminating concerns about external data processing.
- Enterprise Compliance. Leverage Azure's comprehensive compliance certifications and security standards, with Azure serving as your primary cloud provider for all processing activities.
- Simplified Procurement. Use your existing Azure billing relationship and enterprise agreements, consolidating costs and eliminating the need for separate vendor management.
- Cost Structure. Azure managed applications use consumption-based pricing, combining the underlying Azure compute costs with the Upstage service fees, billed directly through your Azure account.
Prerequisites
Before deploying an Upstage product on Azure, ensure you have the following requirements:
- Active Azure Subscription - with administrative privileges to create resources and managed applications.
- Resource Quotas - Adequate GPU compute quotas in your target Azure region to support the service requirements.
- Network Infrastructure - An existing virtual network that can establish VPC peering with the virtual network created by the managed application for secure connectivity.
Managing Azure Resource Quotas
This section guides you through verifying and adjusting your Azure compute quotas. Skip ahead if you're confident your quotas are already sufficient.
The GPU VM family you need depends on the product you're deploying:
| Product | VM family vCPU quota to check |
|---|---|
| Information Extract | Standard NVADSA10v5 Family vCPUs, Standard NCadsH100v5 Family vCPUs |
| Document Parse | Standard NVADSA10v5 Family vCPUs |
| Studio | Standard NCadsH100v5 Family vCPUs |
Verify Current Quotas
- Navigate to Subscriptions in the Azure portal.
- Choose your target subscription and click Usage + quotas in the Settings panel.
- Search for the VM family vCPU quota that matches the product you're deploying (see the table above) and filter by your deployment region.
- Confirm you have adequate quota for your planned deployments. Each service has different requirements - if your current limits are insufficient, proceed to request additional quota.
Request Additional Quota
To increase your compute quotas:
- Select the relevant VM family vCPU quota checkbox (per the table above) and choose New Quota Request > Enter a new limit.
- Specify your required quota amount in the New limit field and submit the request.
Deployment
Expand the product you want to deploy. Each runs as its own Azure managed application.
Cleanup and Resource Management
Remove Deployed Services
When you no longer need a deployment:
- Navigate to your managed application in the Azure Portal.
- Click Delete in the overview section and confirm the action.
- Azure automatically handles cleanup of the associated managed resource group - no additional action required.

Common Questions & Errors
For additional assistance with Azure marketplace deployments or technical support, please reach out to Upstage Support.
Sagemaker ML Products
Deploy Solar Pro 2, Solar Mini, Document Parse, Document OCR, Information Extract, and Embedding as SageMaker endpoints in your own AWS account, with real-time and batch transform examples.
Snowflake Native App
Run Upstage Document Parse inside your Snowflake account as a serverless Native App β prerequisites, installation, usage from a stage or base64, and cost management tips.