AI at JGU

The ZDV provides an AI service for JGU. This includes various large language models.

  1. Log in
  2. Technical Basics
    1. Available AI Models
    2. System Components
  3. In Process of Development
  4. Support
  5. More Information

Log in

You will need a JGU account to log in. The AI chat is initially available to employees. The offer for students is currently being prepared.
Log in at https://ki-chat.uni-mainz.de with the login details of your JGU account.

To Top of Page

Technical Basics

The ZDV controls and hosts the service itself, so the data (except for the web search) remains within JGU.

To Top of Page

Available AI Models

AI Model Description Specifications
Nemotron Ultra 253B Strong performance in all areas, optional reasoning for improved logic capabilities Derived by Nvidia from Llama 3.1 405B,
context window: 128k tokens, knowledge cutoff: December 2023
Gemma3 27B Multimodal processing of text & images, trained on over 140 languages, very efficient Google's open counterpart to Gemini, context window: ~90k tokens (with us), knowledge cutoff: August 2024
Qwen2.5 Coder 32B Specialized in programming tasks, accurate & fast coding at the level of the best models without reasoning Polyglot coder from Alibaba, context window: 32k tokens, knowledge cutoff: June 2024
To Top of Page

System Components

  1. Frontend (OpenWebUI): User interface for interaction.
  2. Load balancer (LiteLLM): Central interface to AI server(s).
  3. AI server (vllm): Executes the AI models on the GPUs.
To Top of Page

In Process of Development

  • Preparations are currently underway to offer the service to students
  • Improved extraction and search in documents
  • Creation of permanent “knowledge repositories”
  • Connection of applications such as Moodle or BBB
  • First tool calling integrations (e.g. calculators)
  • Regular update of the AI models
To Top of Page

Support

Information and training materials are currently being prepared for use in collaboration with HE. Please send your current support requests to digitale-lehre@uni-mainz.de.

To Top of Page

More Information

The Digital Teaching Competence Team has created a website on the topic of AI in higher education. There you will find further information and event information:
https://digitale-lehre.uni-mainz.de/ki-in-der-hochschulbildung/ (german only)

To Top of Page