ChatGPT Edu
Access Instructions
ChatGPT for Education is a secure, privacy-focused version of OpenAI’s advanced AI tool, designed for responsible use with university data. As a leading provider of generative AI, OpenAI ensures that no data entered in the Duke ChatGPT Edu workspace is used to train its models.
Currently this service is for Duke University affiliations only. It has not been approved for health system use, or PHI. Duke Health users should continue to use Microsoft CoPilot for AI services at this time.
ChatGPT Edu can answer questions, write, edit, synthesize data and ideas, and even inspire - increasing productivity, learning and creativity. It includes:
- Access to GPT‑4o, ChatGPT Edu's flagship model, excelling in text interpretation, coding, and mathematics
- Robust security and data privacy, allowing for use of Duke’s sensitive and protected data (but not PHI). Learn more about Duke data classifications.
- Advanced capabilities such as data analytics, web browsing, and document summarization
- The ability to build GPTs, custom versions of ChatGPT, and share them
- Significantly higher message limits than the free version of ChatGPT
- Improved language capabilities across quality and speed, with over 50 languages supported
At this time the Standard license does not include free access to advanced reasoning models, voice mode, deep research or tasks.
Important: Duke users’ conversations and data will not be used to train OpenAI models.
Who Has Access to ChatGPT Edu
All undergraduate students at Duke will have unlimited access to ChatGPT-4o at no cost. Current ChatGPT users of the free and “Plus” versions will have the ability to migrate into Duke’s secure licensed agreement.
Additionally, staff, faculty, and graduate and professional students in the Nicholas School of the Environment, the Sanford School of Public Policy, Pratt School of Engineering, Fuqua School of Business, Learning Innovation and Lifetime Education, and the Office of Information Technology will also be able to access the license without cost.
All other members of the University Community will be able to purchase the license for reduced cost.
Additional tiers with cost premiums based on usage (fund code required). Advanced and premium access will soon be available for undergraduate and graduate students.
-
Advanced - Standard features plus o4 mini, o4 - mini-high, native image gen, GPT-4.5, o3, 4.1, 4.1 mini, Advanced Voice
-
Premium* - (complex research tools) Standard and Advanced features plus o1-Pro, Deep Research
* Premium models, including Deep Research and o1-pro, consume significantly more credits and may result in higher costs charged to your fund code. To avoid unnecessary expenses, use these models only when their advanced capabilities are required.
- Go to https://software-manager.oit.duke.edu/
- After placing your order, you will receive an invite from OpenAI to join Duke’s workspace
- Go to https://chatgpt.com to log in
- Must log in as NetID@duke.edu for account setup
- Once provisioned you will be redirected to Duke SSO page for login
* Important note: If you previously created a personal or consumer account using your NetID@duke.edu email address, please be aware that signing up for a Duke-managed account may require migrating your existing data.
OpenAI Academy
The following links are provided by OpenAI to help you become familiar with ChatGPT Edu.
Frequently Asked Questions
How is ChatGPT Edu different than ChatGPT?
ChatGPT Edu is designed specifically for educational institutions, offering a secure, private, and institutionally managed version of ChatGPT. ChatGPT Edu does not train on user data, ensuring your interactions remain confidential. It provides access to more advanced, high performance models that have more reliable responses and higher usage limits than the ChatGPT free plan.
What is the difference between model names like GPT-4o and versions such as GPT-4.1 or 4.5?
While the numeric versions may sound newer, GPT-4o is currently the most advanced and capable general-purpose model available from OpenAI. It's faster, more efficient, and often more accurate than previous GPT-4 variants, including 4.1, while also enabling rich multimodal interactions (like real-time voice and image understanding). In short, GPT-4o is not a stripped-down version — it's a more powerful, next-generation model built for a wider range of uses.
What does the “o” in GPT-4o stand for?
It stands for Omni, reflecting the model's unique ability to handle text, vision, and audio natively.