# Can Small Models Reason About Legal Documents? A Comparative Study

arXiv: https://arxiv.org/abs/2603.25944 · HTML: https://mlsanity.com/paper/2603.25944/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning.
- Builds on 11 of its 38 references that are in the map.
- First posted on arXiv 26 Mar 2026.

## Placed in

- LLM reasoning / [Reasoning for agents, planning and code](https://mlsanity.com/lr/reasoning-for-agents-planning-and-code/)

## Builds on (most cited first, 11 of 11)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [GPT-4o System Card](https://mlsanity.com/paper/2410.21276/) (2024)
- [Mistral 7B](https://mlsanity.com/paper/2310.06825/) (2023)
- [Gemma 3 Technical Report](https://mlsanity.com/paper/2503.19786/) (2025)
- [To CoT or not to CoT? Chain-of-thought helps mainly on math and symbolic reasoning](https://mlsanity.com/paper/2409.12183/) (2024)
- [SaulLM-7B: A pioneering Large Language Model for Law](https://mlsanity.com/paper/2403.03883/) (2024)
- [LegalBench-RAG: A Benchmark for Retrieval-Augmented Generation in the Legal Domain](https://mlsanity.com/paper/2408.10343/) (2024)
- [ChatGPT may Pass the Bar Exam soon, but has a Long Way to Go for the LexGLUE benchmark](https://mlsanity.com/paper/2304.12202/) (2023)
- [LRAGE: Legal Retrieval Augmented Generation Evaluation Tool](https://mlsanity.com/paper/2504.01840/) (2025)

## Abstract (arXiv)

Large language models show promise for legal applications, but deploying frontier models raises concerns about cost, latency, and data privacy. We evaluate whether sub-10B parameter models can serve as practical alternatives by testing nine models across three legal benchmarks (ContractNLI, CaseHOLD, and ECtHR) using five prompting strategies (direct, chain-of-thought, few-shot, BM25 RAG, and dense RAG). Across 405 experiments with three random seeds per configuration, we find that a Mixture-of-Experts model activating only 3B parameters matches GPT-4o-mini in mean accuracy while surpassing it on legal holding identification, and that architecture and training quality matter more than raw parameter count. Our largest model (9B parameters) performs worst overall. Chain-of-thought prompting proves sharply task-dependent, improving contract entailment but degrading multiple-choice legal reasoning, while few-shot prompting emerges as the most consistently effective strategy. Comparing BM25 and dense retrieval for RAG, we find near-identical results, suggesting the bottleneck lies in the language model's utilization of retrieved context rather than retrieval quality. All experiments were conducted via cloud inference APIs at a total cost of 62, demonstrating that rigorous LLM evaluation is accessible without dedicated GPU infrastructure.