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

TL;DR

SteuerLLM is a specialized large language model for German tax law that outperforms general models by training on domain-specific synthetic data and using a structured evaluation benchmark, advancing legal AI capabilities.

Contribution

The paper introduces SteuerEx, a novel benchmark for German tax law, and SteuerLLM, a domain-adapted LLM trained on synthetic data, demonstrating improved performance over general models.

Findings

SteuerLLM outperforms comparable general-purpose models.

Domain-specific training data enhances legal reasoning accuracy.

Open release of datasets and models supports reproducible research.

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.