SteuerLLM: local specialized large language model for German tax law analysis

Scientific Reports, vol. 16

Abstract

Large language models (LLMs) demonstrate strong general reasoning and language understanding, yet their performance degrades in domains governed by strict formal rules, precise terminology, and legally binding structure. Tax law exemplifies these challenges, as correct answers require exact statutory citation, structured legal argumentation, and numerical accuracy under rigid grading schemes. We curate SteuerEx, the first open benchmark derived from authentic German university tax-law examinations, processed through a structured statement-level decomposition pipeline. SteuerEx comprises 115 expert-validated examination questions spanning six core tax law domains and multiple academic levels, and employs a statement-level, partial-credit evaluation framework that closely mirrors real examination practice. We further present SteuerLLM, a domain-adapted LLM for German tax law trained on a large-scale synthetic dataset generated from authentic examination material using a controlled retrieval-augmented pipeline. SteuerLLM (28B parameters) consistently outperforms general-purpose instruction-tuned models of comparable size and, in several cases, substantially larger systems, demonstrating that domain-specific data and architectural adaptation are more decisive than parameter count for performance on realistic legal reasoning tasks. All benchmark data, training datasets, model weights, and evaluation code are released openly to support reproducible research in domain-specific legal artificial intelligence. A web-based demo of SteuerLLM is available at https://steuerllm.i5.ai.fau.de .

Authors 11

1. Sebastian Wind corresponding

   Friedrich-Alexander-Universität Erlangen-Nürnberg · DATEV (Germany)

   Affiliation as printed

   DATEV eG, Nuremberg, Germany. sebastian.wind@fau.de

   Erlangen National High Performance Computing Center, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany. sebastian.wind@fau.de

   Pattern Recognition Lab, Friedrich-Alexander-Universität Erlangen-Nürnberg, Martensstr. 3, 91058, Erlangen, Germany. sebastian.wind@fau.de

2. Jeta Sopa

   Friedrich-Alexander-Universität Erlangen-Nürnberg

   Affiliation as printed

   Pattern Recognition Lab, Friedrich-Alexander-Universität Erlangen-Nürnberg, Martensstr. 3, 91058, Erlangen, Germany

3. Laurin Schmid

   Friedrich-Alexander-Universität Erlangen-Nürnberg · University of Technology Nuremberg

   Affiliation as printed

   Bavarian AI Taxation Laboratory, Department of Computer Science, University of Technology Nuremberg, Nuremberg, Germany

   Pattern Recognition Lab, Friedrich-Alexander-Universität Erlangen-Nürnberg, Martensstr. 3, 91058, Erlangen, Germany

4. Quirin Jackl

   Friedrich-Alexander-Universität Erlangen-Nürnberg

   Affiliation as printed

   Chair for Tax Law and Public Law, Friedrich-Alexander-Universität Erlangen-Nürnberg, Nuremberg, Germany

5. Sebastian Kiefer

   DATEV (Germany)

   Affiliation as printed

   DATEV eG, Nuremberg, Germany

6. Fei Wu

   Friedrich-Alexander-Universität Erlangen-Nürnberg

   Affiliation as printed

   Pattern Recognition Lab, Friedrich-Alexander-Universität Erlangen-Nürnberg, Martensstr. 3, 91058, Erlangen, Germany

7. Martin Mayr

   Friedrich-Alexander-Universität Erlangen-Nürnberg

   Affiliation as printed

   Erlangen National High Performance Computing Center, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany

   Pattern Recognition Lab, Friedrich-Alexander-Universität Erlangen-Nürnberg, Martensstr. 3, 91058, Erlangen, Germany

8. Harald Köstler

   Friedrich-Alexander-Universität Erlangen-Nürnberg

   Affiliation as printed

   Chair of Computer Science 10, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany

   Erlangen National High Performance Computing Center, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany

9. Gerhard Wellein

   Friedrich-Alexander-Universität Erlangen-Nürnberg

   Affiliation as printed

   Erlangen National High Performance Computing Center, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany

10. Andreas Maier

    Friedrich-Alexander-Universität Erlangen-Nürnberg

    Affiliation as printed

    Erlangen National High Performance Computing Center, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany

    Pattern Recognition Lab, Friedrich-Alexander-Universität Erlangen-Nürnberg, Martensstr. 3, 91058, Erlangen, Germany

11. Soroosh Tayebi Arasteh Aachen Department of Diagnostic and Interventional Radiology Lab for AI in Medicine

    RWTH Aachen University · Universitätsklinikum Aachen · Friedrich-Alexander-Universität Erlangen-Nürnberg

    Affiliation as printed

    Department of Diagnostic and Interventional Radiology, University Hospital RWTH Aachen, Aachen, Germany

    Lab for AI in Medicine, RWTH Aachen University, Aachen, Germany

    Pattern Recognition Lab, Friedrich-Alexander-Universität Erlangen-Nürnberg, Martensstr. 3, 91058, Erlangen, Germany