The AI Innovation Hub offers access to an AI Platform service that allows the Cornell community to securely and privately access many popular AI models, such as large language models (LLMs), image/text embedding models, text-to-image models, and more. The AI Platform is self-hosted on the Cornell cloud infrastructure.

AI Gateway

The AI Gateway gives you safe, governed access to AI models as well as secure tools and connectors through the AI Platform service. This allows you to use LLMs on your behalf with university systems of record in a secure Cornell environment. With the AI Gateway you can access frontier models such as Google Gemini, Anthropic Claude, OpenAI GPT, and experiment with other models like Mistral, AWS Nova and Grok. It also enables you to power tools like Claude Code and OpenWebUI, and to seamlessly integrate AI into your custom scripts and application development.

AI Agent Studio

The AI Platform also gives access to the Agent Studio, providing a safe, governed space for building an automation environment for IT staff, and those fluent in AI coding. With the Agent Studio you can create your own AI agents and automations. Builders work conversationally through Claude Cowork or Claude Code, in Claude Desktop; the platform does the heavy lifting behind them.

Data Risk Levels

If you are unsure what risk category your data falls into, refer to University Policy 5.10 for definitions of Low/Moderate/High Risk data and what types of data fall into each category.

Low and Moderate Risk Data

Moderate Risk and Low Risk data is permissible in the AI Platform. If you plan to use FERPA data within the API, please inform us of your intentions. 

High Risk Data

If you are working with High Risk data and would like to utilize the AI Platform with that data, please reach out to us. We will discuss your use-case with the IT Security Office and find a path forward.

Cost

To use the AI Platform, each team must provide a KFS account number, which will be associated with their token consumption, and usage will be charged to this account on a monthly basis. As this process is new, there may be some initial delays in billing setup. Additionally, we offer a monthly free tier of $10 to allow new users to explore and test the service at no cost up to this limit. The free tier applies at the team level, not individual api keys.

To help manage expenses, we recommend that Team admins set Budgets and Rate Limits for each API Key they create.

View the available models and associated pricing on the AI Platform Confluence page (requires a login).