Our Principal Investigator Frank Hutter, leading the AutoML group at the ELLIS Institute Tübingen since May 2024, and professor at University of Freiburg, has received the KI-Innovationspreis at this year’s Deutscher KI-Preis (German AI Award) for his long-term work on AutoML and the recent TabPFN (Tabular Prior-data Fitted Network) he developed with his team. Awarded by WELT (Axel Springer), the innovation prize recognises outstanding scientific contributions to artificial intelligence and is considered the most highly endowed award of its kind in Germany and Europe.
TabPFN is transforming how we work with tabular data. The machine-learning algorithm learns causal relationships from synthetic data and delivers highly precise predictions on small datasets - a capability that is decisive for fields such as medicine, supply chains, and financial reporting. Trained once on 100 million synthetic datasets, it makes accurate predictions on new tables of up to around 10,000 rows without any task-specific tuning, matching or surpassing the strongest conventional methods in a fraction of the time; seconds rather than hours. The work was published in Nature in 2025 and has since been widely taken up by the research community.
The reach of TabPFN now extends well beyond academia. Frank co-founded the spin-off Prior Labs in 2024 to bring the model into real-world use, and SAP has acquired the company to build a frontier AI lab for structured data.
The award places Frank in a distinguished line of Tübingen scientists. Bernhard Schölkopf, scientific director of the ELLIS Institute Tübingen, received the innovation prize in 2020, and the ELLIS network itself was honoured in 2021. Frank, who also directs the ELLIS unit Freiburg and has co-founded Prior Labs to bring TabPFN to real-world applications, is a Fellow of ELLIS and EurAI and a three-time ERC grantee.
The ELLIS Institute Tübingen warmly congratulates Frank Hutter and his team on this well-deserved recognition.