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

2026-02-11

AI Summary

This study addresses the limitations of general-purpose large language models in highly structured and regulation-intensive domains such as German tax law, where precise statutory citation, logical reasoning, and numerical computation are critical. To bridge this gap, the authors introduce SteuerEx, the first fine-grained open benchmark based on authentic German university tax law exam questions, and propose a domain-adaptation training approach that integrates retrieval-augmented generation with controlled synthetic data. The resulting specialized model, SteuerLLM, significantly outperforms same-sized general models across multiple tax-related tasks and even surpasses larger-scale systems. Evaluated in realistic examination settings, SteuerLLM demonstrates high accuracy and practical utility, underscoring that domain specialization is more effective than merely scaling up model size.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Application Domains: Humanities & Computational Social Science

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

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 algorithmically generate SteuerEx, the first open benchmark derived from authentic German university tax law examinations. 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 scale 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.

Problem

• legal reasoning
• tax law
• domain-specific LLM
• formal rules
• statutory citation

Innovation

• domain-adapted LLM
• legal reasoning benchmark
• synthetic data generation
• retrieval-augmented pipeline
• tax law analysis