From $1.50 usage-based · Free plan🇮🇱
Web data provider with proxy networks, scraper APIs, a hosted scraping browser and an AI tool that generates scrapers.
LegalBench overview and key facts (October 2026)
Collaborative benchmark of 162 English legal reasoning tasks from 40 contributors, covering contracts, statutes and court opinions.
Free
What is LegalBench?
LegalBench is an open benchmark for evaluating legal reasoning in English language models. Legal and IT teams use it to see how well a model handles contracts, statutes and court decisions.
LegalBench is an open science project that collects tasks from legal professionals and researchers. Its 162 tasks span areas such as evidence, contracts and civil procedure, and many are yes-or-no questions about whether a text has a legal attribute. Each task has a small training split and an evaluation split. Hazy Research at Stanford hosts the project website and GitHub repository; the paper was published in 2023.
LegalBench AI features and use cases
- Evaluates models for legal workBefore a legal team relies on a model to review contracts or summarize rulings, it can test candidates on the tasks closest to its work. Weak spots show up per task type.
- Tests contract clause recognitionSeveral tasks ask whether a contract passage contains a certain clause type. This mirrors routine review work and shows how often a model misses or invents clauses. The results show where human review remains necessary.
- Benchmarks prompt strategiesThe few-shot training splits let teams compare prompts and examples on the same tasks before rolling out a legal assistant. Small changes in prompts can shift results on legal tasks, and the shared tasks make those effects measurable.
What sets LegalBench apart
- Tasks contributed by legal professionals and researchers, grouped by legal area and task type.
- Searchable task descriptions on the project website.
Where LegalBench is strongest
United States Available internationally
LegalBench is run by the Hazy Research group at Stanford University, and its tasks are based on English-language, mostly US legal material.
Questions to ask before you buy LegalBench
All tasks are in English and largely based on US law, so results do not transfer to German or EU law. Check the license of each task you use, since sources differ. Model outputs on legal questions still need review by qualified lawyers.
LegalBench reviews
No reviews yet. Be the first to share how your team uses it.
Every review shows its verification level. We read each one before it goes live. How we check reviews
Questions about LegalBench
Does LegalBench cover German law?
No. All tasks are in English, and most draw on US legal material such as statutes, judicial opinions and contracts.
How many tasks are there?
The benchmark currently has 162 tasks from 40 contributors, covering classification, extraction, generation and entailment.
Where are the task descriptions?
The project website lets you search task descriptions, and the GitHub repository has more detailed descriptions per task.
LegalBench update historyno updates yet
No changes recorded for LegalBench yet.
Changes the vendor made and confirmed user reports, each checked against the vendor's own pages. We review every tool monthly.
Alternatives to LegalBench
Tools in the same category (Domain datasets) with the most overlapping markets.
Free🇫🇷
Open pretraining corpus of 2.27 trillion tokens from Pleias in Paris, containing only uncopyrighted or freely licensed documents.
FreeAvailable internationally
Contract Understanding Atticus Dataset: 510 commercial contracts with 13,000+ lawyer-supervised labels for 41 clause types.
FreeAvailable internationally
Open-book question answering test on public company filings from Patronus AI, with an open sample of 150 annotated cases.