LegalBench: 162 Tasks for Evaluating Legal Reasoning in LLMs

LegalBench is an open science project curating tasks for evaluating legal reasoning in English large language models. The benchmark currently consists of 162 tasks gathered from 40 sources. The dataset is authored by nguha and was last updated on March 30, 2026.

Use Cases

• Benchmarking LLM performance on legal reasoning tasks based on the 162 curated tasks.
• Developing specialized legal AI models based on the structured evaluation framework.
• Studying the capabilities and limitations of general-purpose LLMs in the legal domain.

Strengths

• Consists of 162 distinct tasks for evaluation.
• Collaboratively curated from 40 sources, suggesting a broad scope.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last updated 2026-03-30 05:09:56; freshness should be verified.

Provenance

Source

HazyResearch at Stanford University

Collection Method

Collaborative curation from multiple sources.

Freshness

Last updated March 30, 2026.