# Legal Reasoning on LegalBench (Comprehensive Metrics)

Source: [Legal Reasoning on LegalBench (Comprehensive Metrics)](<https://www.sota2.com/research/sota/legal-reasoning-on-legalbench-comprehensive-metrics>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-06-04T10:17:14.032Z

Evaluation of model performance on Legal Reasoning tasks using the LegalBench dataset, measured across multiple specialized metrics including Accuracy, Decision Accuracy, and various component scores (CR, CD, CS, cMFG variants).

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Legal Reasoning |
| Dataset (including any stated split) | LegalBench |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 2 of 2 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Accuracy](<https://www.sota2.com/research/sota/legal-reasoning-on-legalbench-comprehensive-metrics.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Decision Accuracy](<https://www.sota2.com/research/sota/legal-reasoning-on-legalbench-comprehensive-metrics.md?metric=Decision%20Accuracy>): Higher is better; unit: PERCENT.

- [CR Score](<https://www.sota2.com/research/sota/legal-reasoning-on-legalbench-comprehensive-metrics.md?metric=CR%20Score>): Higher is better; unit: PERCENT.

- [CD Score](<https://www.sota2.com/research/sota/legal-reasoning-on-legalbench-comprehensive-metrics.md?metric=CD%20Score>): Higher is better; unit: PERCENT.

- [CS Score](<https://www.sota2.com/research/sota/legal-reasoning-on-legalbench-comprehensive-metrics.md?metric=CS%20Score>): Higher is better; unit: PERCENT.

- [cMFG (R)](<https://www.sota2.com/research/sota/legal-reasoning-on-legalbench-comprehensive-metrics.md?metric=cMFG%20(R)>): Higher is better; unit: PERCENT.

- [cMFG (D)](<https://www.sota2.com/research/sota/legal-reasoning-on-legalbench-comprehensive-metrics.md?metric=cMFG%20(D)>): Higher is better; unit: PERCENT.

- [cMFG (S)](<https://www.sota2.com/research/sota/legal-reasoning-on-legalbench-comprehensive-metrics.md?metric=cMFG%20(S)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| QwQ-32B | 82.3 | Not recorded | 6a210696c10484bac23f021d |
| DeepSeek-R1-8B | 76.2 | Not recorded | 6a210696c10484bac23f021d |

### QwQ-32B

Accuracy: 82.3. Result ID: 6a21509afbc373592c1d96c1.

[Reporting paper](<https://www.sota2.com/research/paper/6a210696c10484bac23f021d.md>)

Recorded evaluation configuration: {"Model Type":"Natively-trained reasoning LM","Parameters":"32B"}. This record does not assert independent reproduction.

### DeepSeek-R1-8B

Accuracy: 76.2. Result ID: 6a21509afbc373592c1d96c0.

[Reporting paper](<https://www.sota2.com/research/paper/6a210696c10484bac23f021d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76a91e209131cea6e3eb.md>)

Recorded evaluation configuration: {"Model Type":"Distilled LRM","Parameters":"8B"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.
