Build secure AI apps with managed OpenAI tools

HMS Azure AI provides secure, project-specific access to enterprise-grade OpenAI capabilities in a secure environment. Teams can build intelligent applications using managed OpenAI models, the Assistant API with vector search, and global batch deployment options. It is designed for scalability, flexibility, and efficiency. This service is available to licensed project developers and projects needing to safeguard data classified as Level 4 and below according to the University risk classifications.

Learn more about HMS Azure AI

Eligibility

Restricted to licensed project developers.
Faculty Staff

Access to HMS Azure AI is limited to HMS Quad-based faculty and staff with project-specific (data security Level 4) security requirements.

To request access, email hms-it-ai@hms.harvard.edu with a brief description of your project and intended use of the environment.

Security

Meets data security level 4 security with API key access.
Level 4 API key

This service is designed for secure use cases up to data security Level 4. There are no specific network restrictions to connect; users authenticate directly via their assigned API key. The user is responsible for safeguarding this key and using it appropriately.

Features

Enterprise-grade tools for scalable AI development.
Models APIs

HMS Azure AI offers a robust set of features to support research and development:

  • OpenAI models as a service Use enterprise-grade OpenAI models through a managed, secure interface.
  • Assistant API and vector search – Build conversational AI applications with vector-based semantic search capabilities.
  • Global batch deployment models – Handle large-scale data processing across global regions with parallel execution and significant cost savings.
  • Access to all API versions – Utilize all available OpenAI API versions to remain current with new capabilities.

Cost

Uses a token-based billing model.
Token-based Usage

HMS Azure AI uses a token-based usage model, allowing users to scale resources according to their project needs and budget.