GPTalk: Major Update with New Features  [10.04.26]

GPTalk, the University of Hohenheim’s AI chat interface, has received a major update. Here’s an overview of the most important new features.

Process files directly in the chat

GPTalk can now process files. Documents, spreadsheets, presentations, code, or images can be attached directly to a message and analyzed, summarized, or further processed by the AI. Common formats such as PDF, Word, Excel, PowerPoint, CSV, image files, and many more are supported—with a maximum file size of 10 MB.

Please note: Only files that you have created yourself or for which there is a legal basis for use in a university context may be uploaded. Users are responsible for ensuring compliance with copyright laws. All uploaded files will be automatically deleted after one month at the latest.

Web Search

Selected OpenAI models can now search the internet natively, allowing them to incorporate up-to-date information into their responses.

New Default Model

The new default model is OpenAI GPT 5.4 mini—fast, powerful, and ideally suited for most tasks. In addition, a wide range of models, some of which are new, remains available.

Significantly Better Response Quality

Thanks to a fundamental overhaul of the system prompt, GPTalk now delivers noticeably more precise, structured, and useful responses—across all available models.

Control Creativity Yourself

A new setting lets you choose whether the AI should respond in a precise, balanced, or creative manner. More creative responses can be more inspiring, but they also increase the risk of inaccuracies.

Intelligent model filtering

When you attach files or enable web search, you’ll automatically see only the models that support these features. No more guessing which model can do what.

Cost display and improved presentation

After each response, you’ll now see how many tokens were processed and what the cost was. Additionally, text display, URL previews, and accessibility have been improved.

Back to All news