Starting with the 2026/2027 winter semester, the university's AI chat receives several new updates. These are active immediately, allowing you to test the new functions and the new model.
New Web Interface
The web interface is changing with the new OpenWebUI version 0.11.4. The model selection drop-down menu is now located at the bottom right of the chat window.
Login here: https://ki-chat.uni-mainz.de
Model Update
The new DeepSeek V4.1 Flash model replaces MiniMax M3. You can use it for general chat inquiries as well as complex and agentic tasks. Additionally, you can use it for coding, generating and summarizing text, and analyzing images and documents. The following models are available to you for the 2026/2027 winter semester: the new DeepSeek V4.1 Flash model, the previously introduced Qwen3.6 model, and the two image models Z-Image-Turbo and Qwen-Image-Edit-2511 Lightning variant.
New Features: Transcription and Text Extraction
Two new features transcription and text extraction are accessible via the model selection drop-down menu. With the transcription feature, the AI chat converts audio into text (e.g., from interviews, lectures, meetings, or voice memos). You can then process this text further within the same chat, asking the AI to summarize, translate, or rephrase it. With the text extraction feature, the AI chat converts scanned documents, photos of whiteboards, screenshots, and forms into text.
Higher Token Limits
The AI chat now allows for a higher token limit, text units, you can use within a given timeframe. These limits may be adjusted based on system load. If a request reaches the limit, the AI chat will display the currently applicable limit in the error message.
Older models discontinued
As announced, the ZDV discontinued the GPT OSS 120B, Qwen3 235B, and Qwen3 Coder 30B models in early October 2026.
For more information about the models and other important notes, please visit our website:
https://www.en-zdv.uni-mainz.de/ai-at-jgu/
More news from the Data Center → may be found here.