The OTDI Cloud Services Group, in partnership with GCP, AWS, Azure, and the Ohio Supercomputer Center (OSC), provides Ohio State researchers, professors, and developers with streamlined, programmatic access to Large Language Models (LLMs) that have been vetted by the university to ensure institutional data protection and IDP Policy Compliance.
Please Note: This service is built specifically for developers and builders. Model access is solely provided through API calls for integration into your custom applications, scripts, and supported tools. There is no graphical chat interface.
Data Security and Classification
Because this service is routed through our secured endpoint, it’s been approved for use with all Ohio State institutional data classifications, from S1 (Public) to S4 (Restricted). However, we still suggest you check with your organization’s security coordinator before starting any project.
Request a GenAI Account
Learn how to request a Generative AI account through Workday and ServiceNow.
Supported Providers and Model Families
Through our endpoint, you’ll have access to industry-leading models from most major cloud providers. Highlights include:
- Azure (OpenAI): GPT models, including flagship and lightweight tiers, alongside advanced reasoning models.
- Vertex AI (Google): Gemini models across both Pro and Flash tiers.
- AWS Bedrock: Anthropic Claude (including Sonnet, Opus, and Haiku tiers), Amazon Nova, Meta Llama, DeepSeek, and Qwen models.
- OSC: Coming Soon — expanded access to open-weight models.
For a complete, real-time list of every specific model version, limits, capabilities and token costs, please review our Model Hub Table.
How It Works
You access the available LLMs by routing your requests through our custom endpoint using a Virtual API key. Because our endpoint uses the standard OpenAI protocol, any AI tool, library, or application that allows you to define a custom OpenAI BaseURL or Endpoint will work seamlessly with our service.
Billing and Team Management
This service operates on a ‘Chargeback’ model, meaning your billing account will be invoiced, one month in arrears, based strictly on your token consumption. The billing account is defined by a Workday Purchase Requisition.
- Teams and Admins: Each billing entity is assigned a dedicated Team with a maximum budget cap and designated Team Admin(s).
- API Key Controls: Team Admins have full control over access. They can generate custom API keys for users and apply granular restrictions based on:
- Date and time limits
- Maximum budget caps
- Specific model access
Common Integrations and Use Cases
Our users are currently leveraging this custom endpoint across a wide variety of workflows, including:
- Custom Code: Python scripts for academic and institutional research.
- Notebooks: Google Colab, Claude Code with CoWork, etc.
- Command Line Interfaces and Development Tools: Claude Code CLI, Codex CLI, Open Code CLI, and more. (See our guide for Installing GenAI Command Line Interfaces.)
To Get Started
Request a Generative AI Service Account
This is a two-step process. First, you must set up the billing account with a Purchase Requisition in Workday. Once your Purchase Requisition is approved and converted to a Purchase Order, you’ll be able to complete the second part: Requesting a full production account through a ServiceNow Catalog Request Form. How to Request a Generative AI Account will walk you through the entire process.
