Awards, papers, releases, media and events. Every notable milestone since 2009.
FILTER
YEAR
Sep
At FACIS Connect '26, Prof. Dr. Alexander Löser joined industry leaders from Airbus, Schaeffler, and the International Manufacturing-X Council for the panel discussion "Industrial Collaboration in the Digital Age", focusing on European digital sovereignty and the SOOFI project.
Sep
Tom Röhr presented SOOFI, a sovereign open-source German foundation model, at DevDay — covering its hybrid Mixture-of-Experts + Mamba-2 architecture, trained on 512 GPUs across 64 nodes over 51 days by an 11-partner consortium including Merantix Momentum. A live demo showed the model writing and self-correcting a Conway's Game of Life implementation. Röhr also joined a panel discussion on sovereign AI at the same event.
Sep
Tom Röhr delivered the opening keynote at Haystack Europe 2026 (https://haystackconf.com), the Search Relevance Conference, over two days. The talk built on themes from an earlier Merantix presentation and included a dedicated section on the specialization of LLMs.
Sep
The paper "Same Meaning, Different Scores: Lexical and Syntactic Sensitivity in LLM Evaluation" by Bogdan Kostić, Julian Risch, and Conor Fallon was accepted at LREC 2026.
Sep
Alexander Löser was quoted as an expert in the L'Express article "AI: From the Black Forest to Siemens – Germany's strategic awakening."
Sep
The paper "CliniBench: A Clinical Outcome Prediction Benchmark for Generative and Encoder-Based Language Models" by Paul Grundmann, Dennis Fast, Jan Frick, Thomas Steffek, Felix Gers, Wolfgang Nejdl, and Alexander Löser was accepted as a full paper at EACL 2026.
Sep
The paper "Robust Conformal Prediction for Infrequent Classes" by Jens-Michalis Papaioannou, Sebastian Jäger, Alexei Figueroa, David Stutz, Betty van Aken, Keno Bressem, Wolfgang Nejdl, Felix Biessmann, and Felix Gers was accepted at Transactions on Machine Learning Research (TMLR), joint work with Google DeepMind, L3S, and Grammarly.
Sep
The Soofi S pretraining technical report was published as a Hugging Face Space, documenting the sovereign German-English foundation model's development.
Sep
Tom Röhr spoke at DFKI about multilingual training datasets used for the Soofi model.
Sep
Alexander Löser participated in a panel discussing the sovereignty of AI intelligence made and consumed in the EU, in the context of IPCEI-AI and the SOOFI project.
Sep
The S-PROTO model was published on Hugging Face, supporting classification across 1,900 ICD-10 codes from admission notes and retrieval of similar cases.
Sep
The paper "DeepICD-R1: Medical Reasoning through Hierarchical Rewards and Unsupervised Distillation" by Tom Röhr, Thomas Steffek, Roman Teucher, Keno Bressem, Peter Tröger, and Felix Gers was accepted at LREC 2026.
Jul
SOOFI was featured in the ZDF broadcast "KI-Strategie: Kann Deutschland aufholen?", which covered Germany's AI strategy.
Jul
Digitale Souveränität & the Soofi Project Prof. Dr. Alexander Löser discusses the Soofi model and Germany's chance for AI sovereignty.
Jul
Prof. Dr. Alexander Löser on SOOFI and Europe’s role in the global AI race in Podcast Episode 249: “Deutsches KI-Modell SOOFI: Europas Hoffnung im globalen KI-Wettlauf?
Jul
The sovereign German-English foundation model SOOFI was publicly released on Hugging Face.
Jun
Alexander Löser spoke to the German Judges Association (Brandenburg) about AI in Justice and the Soofi project.
Jun
Alexander Löser joined a panel at FACIS Connect 2026 in Berlin discussing Soofi and modular LLMs.
May
Interactive Multi-Prototype Explanations for Clinical Diagnosis Prediction. Conor Fallon.
Feb
Alexander Löser is featured in the Handelsblatt article "Europa ist bei Sprachmodellen wieder im Spiel," discussing Europe's renewed competitiveness in language models.
Nov
DATEXIS participated in the first international workshop on pretraining and posttraining sovereign foundation models, with Tom Röhr leading a session on reasoning post-training.
Jul
Justus Westerhoff, Golzar Atefi, Mario Koddenbrock, Alexei Figueroa, Alexander Löser, Erik Rodner, Felix Alexander Gers. TMLR 2025.
Jun
The paper "Multinational attitudes toward AI in health care and diagnostics among hospital patients" by Felix Busch et al. was accepted at JAMA Network Open.
May
Jan Frick and Dennis Fast presented at the Medizin Informatik Initiative 2025 workshop in Leipzig.
Feb
The paper "Comply: Learning Sentences with Complex Weights inspired by Fruit Fly Olfaction" by Alexei Figueroa, Justus Westerhoff, Golzar Atefi, Dennis Fast, Benjamin Winter, Felix Alexander Gers, Alexander Löser, and Wolfgang Nejdl was accepted at NICE 2025.
Jan
Alexander Löser contributed a two-page article in Handelsblatt about modular language models and their potential for German industry.
Jan
Alexander Löser appeared on the podcast "Artificial Democracy."
Jan
Prof. Dr. Steffen Staab (University of Stuttgart) and Prof. Dr. Kentaro Inui (MBZUAI) visited DATEXIS.
Jan
Alexander Löser led a workshop with the BMBF Platform Learning Systems on AI, LLMs, policies, and media.
Jan
The paper "'Where does it hurt?' -- Dataset and Study on Physician Intent Trajectories in Doctor Patient Dialogues" by Tom Röhr, Soumyadeep Roy, Fares Al Mohamad, Jens-Michalis Papaioannou, Wolfgang Nejdl, Felix Gers, and Alexander Löser was accepted at the European Conference on AI (ECAI) 2025.
Jan
Alexander Löser gave the laudatio for the Digital Gipfel Startup Award at Connecting the Dots 2025 (BMWK).
Jan
Alexander Löser served as Jury for the BITKOM Startup Award at the AI & Data Summit and Quantum Summit Berlin 2025.
Jan
Alexander Löser joined a Handelsblatt panel on digital business models in Düsseldorf alongside Jörg Bienert (President, KI Verband) and Hanbing Ma (ERGO).
Jan
Alexander Löser was interviewed by Platform Lernende Systeme about how Europe can stay competitive in AI through modular Mixtures of Experts and Large Reasoning Models.
Jan
Alexander Löser appeared on the Social Economy podcast discussing "Artificial Intelligence in the Social Economy – Visions, Opportunities, and Challenges for 2026."
Jan
Alexander Löser was invited as an expert on NFDI Data Infrastructures – Connected & Resilient AI Enablers.
Jan
The paper "Longhealth: A question answering benchmark with long clinical documents" by Lisa Adams, Felix Busch, Tianyu Han, Jean-Baptiste Excoffier, Matthieu Ortala, Alexander Löser, Hugo J.W.L. Aerts, Jakob Nikolas Kather, Daniel Truhn, and Keno Bressem was accepted at the Journal of Healthcare Informatics Research.
Jan
Alexander Löser contributed a two-page Handelsblatt article on modular language models and their potential for German industry.
Jan
The paper "Evaluating Financial Literacy of Large Language Models through Domain Specific Languages for Plain Text Accounting" by Alexei Gustavo Figueroa Rosero, Paul Grundmann, Julius Freidank, Wolfgang Nejdl, and Alexander Löser was accepted at COLING 2025 Workshops.
Jan
The paper "SPONGE: Competing Sparse Language Representations for Effective Cross-Lingual Knowledge Transfer in Healthcare" by Jens-Michalis Papaioannou, Alexei Figueroa, Conor Fallon, Anna Capilla, Alexandra Bekiaridou, Stavros Zanos, Wolfgang Nejdl, and Alexander Löser was accepted at TMLR 2025.
