# Welcome to uniGPT

uniGPT is the University of Münster's AI platform, providing students, researchers, and employees with access to a wide range of large language models (LLMs) — both self-hosted on-premises models and external providers like OpenAI, Google Gemini, and Anthropic Claude.

## What can you do with uniGPT?

- **Chat** with AI models via a modern web interface
- **Access models programmatically** through an OpenAI-compatible API
- **Generate images** using state-of-the-art diffusion models
- **Upload and analyze documents** directly in the chat

## Key Highlights

- **GDPR-compliant**: Internal models run entirely on university infrastructure — your data never leaves the university.
- **No separate subscriptions needed**: Access models from OpenAI, Google, and Anthropic through a single interface without managing individual accounts.
- **Free for all university members**: Available to all students and employees with a university account.

## Getting Started

1. Navigate to [gpt.uni-muenster.de](<https://gpt.uni-muenster.de>)
2. Log in with your university credentials (SSO).
3. Select a model and start chatting!

> [!TIP]
>
> **First time?**
>
> We recommend starting with one of the internal models (e.g., **gemma-4-31B-it**, **DeepSeek V4 Flash**, **medgemma-1.5-4b-it** for medical use cases, or **Qwen3.5-35B-A3B**) — they are fast, free of external cost, and fully GDPR-compliant.

## Support

Need help? Join our support channel: [**uniGPT Support**](<https://mattermost.uni-muenster.de/wwu/channels/unigpt>)

## E-Learning Course

A German-language e-learning course is available for getting started with uniGPT:

[**UniGPT E-Learning (Confluence)**](<https://confluence.uni-muenster.de/courses/playcourse.action?course=346116963>)

The course covers 10 sections across approximately **48 minutes**, including chat settings, model selection, prompting techniques, risk awareness, and a final quiz. After completing it, you will be able to confidently use uniGPT in your daily work and understand the different use cases across various contexts.

## Terms of Use &amp; Privacy

The usage policy and privacy statement can be found directly below the chat input field in the web interface.

## Citing uniGPT

If you use uniGPT in your research, teaching, or other scientific work, please cite the publication describing this service:

> Radas, J., Risse, B., &amp; Vogl, R. (2026). *UniGPT Revisited: From a Simple Chatbot to an API-First AI Platform — Two Years of On-Premises LLM Operations*. In L. Desnos, C. Diaz, J. Mincer-Daszkiewicz, L. Merakos, R. Vogl, &amp; S. U. McLellan (Eds.), *EPiC Series in Computing: Vol. 109. Proceedings of EUNIS 2026 Annual Congress* (pp. 96–107). EasyChair. doi: [10.29007/4rq8](<https://doi.org/10.29007/4rq8>)
