# Legal benchmarks

Find tests for legal reasoning, research, case retrieval, and document work.

Canonical page: https://benchlm.ai/benchmarks/collections/legal

9 records. Jurisdiction, task set, tools, and grading change what each result means. These tests do not establish legal accuracy in a particular matter or replace a review of your workflow.

Records keep metrics and harness variants separate. Source rows can include different model and agent configurations; linked profiles count distinct public model identities. Public model rows link to original published artifacts, which may include publisher comparisons of other providers' models. These counts do not represent independent evaluations. Dates describe the stored source record.

## Legal reasoning and research

| Benchmark | Stored coverage | Task and run conditions | Original source |
|---|---|---|---|
| [PRBench Legal](/benchmarks/scale-prbench-legal) | 41 stored source rows; 22 linked model profiles | Published professional-reasoning configurations remain separate from legal-agent runs. 41 published rows Snapshot label: October 15, 2026 snapshot | [Original source](https://labs.scale.com/leaderboard/prbench-legal) |
| [LegalBench](/benchmarks/valslegalbench) | 151 stored source rows; 85 linked model profiles | The Vals-hosted LegalBench result stays separate from any other LegalBench harness. Legal reasoning task views Source update in snapshot: 2026-10-13; Source snapshot captured: 2026-10-15T18:47:28.403Z | [Original source](https://www.vals.ai/benchmarks/legal_bench) |
| [Legal Research Bench](/benchmarks/legalresearchbench) | 78 stored source rows; 65 linked model profiles | US-law research task views preserve strict and weighted-pass results. US-law legal research tasks Source update in snapshot: 2026-10-12; Source snapshot captured: 2026-10-15T18:47:28.403Z | [Original source](https://www.vals.ai/benchmarks/legal_research) |

## Legal agent work

| Benchmark | Stored coverage | Task and run conditions | Original source |
|---|---|---|---|
| [Harvey's Legal Agent Benchmark](/benchmarks/hlab) | 79 stored source rows; 66 linked model profiles | The Vals-hosted Legal Agent Benchmark preserves task-level and criterion-pass views. Legal agent work across documents, spreadsheets, presentations, and files Source update in snapshot: 2026-10-12; Source snapshot captured: 2026-10-15T18:47:28.403Z | [Original source](https://www.vals.ai/benchmarks/hlab) |
| [LAB all-pass (Anthropic harness)](/benchmarks/legalagentbenchallpass) | 1 public model rows linked to original artifacts | All-pass grading under Anthropic's internal harness. 1,235 legal-agent tasks | [Original source](https://www-cdn.anthropic.com/c5fbac3f0b1280a933ebd26d3cb8bb9f5bdeaf48/Claude%20Opus%205%20System%20Card.pdf) |
| [LAB criterion-pass (Anthropic harness)](/benchmarks/legalagentbenchcriterionpass) | 1 public model rows linked to original artifacts | Criterion-pass grading under Anthropic's internal harness. 1,235 legal-agent tasks | [Original source](https://www-cdn.anthropic.com/c5fbac3f0b1280a933ebd26d3cb8bb9f5bdeaf48/Claude%20Opus%205%20System%20Card.pdf) |
| [LAB all-pass (Harvey held-out)](/benchmarks/legalagentbenchheldoutallpass) | 3 public model rows linked to original artifacts | All-pass grading on the Harvey held-out task set. Harvey-held-out legal-agent tasks | [Original source](https://www-cdn.anthropic.com/c5fbac3f0b1280a933ebd26d3cb8bb9f5bdeaf48/Claude%20Opus%205%20System%20Card.pdf) |
| [LAB criterion-pass (Harvey held-out)](/benchmarks/legalagentbenchheldoutcriterionpass) | 3 public model rows linked to original artifacts | Criterion-pass grading on the Harvey held-out task set. Harvey-held-out legal-agent tasks | [Original source](https://www-cdn.anthropic.com/c5fbac3f0b1280a933ebd26d3cb8bb9f5bdeaf48/Claude%20Opus%205%20System%20Card.pdf) |

## Historical evidence

| Benchmark | Stored coverage | Task and run conditions | Original source |
|---|---|---|---|
| [CaseLaw v2](/benchmarks/valscaselawv2) | 54 stored source rows; 35 linked model profiles | Archived source board. Archived Canadian case-law questions remain useful context for that specific retrieval task. Canadian case-law question answering Source update in snapshot: 2026-05-04; Source snapshot captured: 2026-10-15T18:47:28.403Z | [Original source](https://www.vals.ai/benchmarks/case_law_v2) |

## Other collections

- [Finance benchmarks](/benchmarks/collections/finance)
- [Healthcare benchmarks](/benchmarks/collections/healthcare)

[All benchmarks](/benchmarks)