To content

“Inferenz NRW”: Additional AI models launched for universities in North Rhine-Westphalia

-
in
  • Top-News
A microchip with a holographic map of North Rhine-Westphalia floating above it. © Saskia Molewicz ​/​ Flow
A new service for using language models is now available to universities in North Rhine-Westphalia: Under the project title “Inferenz NRW,” a data protection-compliant alternative to commercial AI services has been developed. The service is aimed at all university members and is designed to meet the rapidly growing demand for AI infrastructure in research, teaching, and administration.

Key features of the new service offering:

  • Versatile use: The models can be used directly as chatbots or integrated into universities’ local applications via API.
  • Focus on open-source and European models: The current portfolio includes GPT OSS (OpenAI), with additional models to be added gradually. The emphasis is on models available for non-commercial use, with a focus on European providers.
  • Infrastructure and expansion: The service currently operates on a “best effort” basis and is subject to certain usage limits due to limited hardware resources. However, an initial expansion of the infrastructure is already planned for the second half of 2026, which will increase capacity. 

The project is backed by a joint working group comprising RWTH Aachen University, the University of Cologne, and Ruhr University Bochum (WestAI, OSKI.nrw, and KI:connect.nrw), supported by the WestAI AI Service Center. Starting in the summer of 2026, the service will be transferred to the KI:Expertisezentrum.nrw project.

Further information and documentation:

For detailed information on the available models, cluster operations, and current usage guidelines, please refer to the official information sheet:

Info PDF on Inferenz NRW 

For more information on TU Dortmund University’s AI offerings, visit:

https://service.tu-dortmund.de/group/intra/ki

The platform itself is accessible at kiconnect.nrw (login for TU members via SSO). Further technical information on the AI models used can be found at oski.nrw.