# Legal Reasoning on LegalBench (accuracy) Source: [Legal Reasoning on LegalBench (accuracy)]() Collection index: [llms.txt]() Data updated: 2026-06-02T10:10:53.005Z Evaluation of model performance on Legal Reasoning tasks using the LegalBench dataset, measured by standard accuracy. 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 | 18 of 18 | | 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](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | evaluation-instructed prompt optimization | 90 | Not recorded | 69268983d6b0dd3342930322 | | Pro-Refine | 86 | Not recorded | 69268983d6b0dd3342930322 | | TextGrad | 84 | Not recorded | 69268983d6b0dd3342930322 | | APE | 84 | Not recorded | 69268983d6b0dd3342930322 | | LLM only | 83 | Not recorded | 69268983d6b0dd3342930322 | | Self-Refine | 81 | Not recorded | 69268983d6b0dd3342930322 | | evaluation-instructed prompt optimization | 70 | Not recorded | 69268983d6b0dd3342930322 | | evaluation-instructed prompt optimization | 69 | Not recorded | 69268983d6b0dd3342930322 | | Self-Refine | 63 | Not recorded | 69268983d6b0dd3342930322 | | Pro-Refine | 63 | Not recorded | 69268983d6b0dd3342930322 | | Self-Refine | 63 | Not recorded | 69268983d6b0dd3342930322 | | Pro-Refine | 63 | Not recorded | 69268983d6b0dd3342930322 | | APE | 61 | Not recorded | 69268983d6b0dd3342930322 | | TextGrad | 58 | Not recorded | 69268983d6b0dd3342930322 | | TextGrad | 58 | Not recorded | 69268983d6b0dd3342930322 | | LLM only | 56 | Not recorded | 69268983d6b0dd3342930322 | | LLM only | 55 | Not recorded | 69268983d6b0dd3342930322 | | APE | 55 | Not recorded | 69268983d6b0dd3342930322 | ### evaluation-instructed prompt optimization Accuracy: 90. Result ID: 6a1eac15fbc373592c19667c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"GPT-4o","Optimization Type":"Query-dependent"}. This record does not assert independent reproduction. ### Pro-Refine Accuracy: 86. Result ID: 6a1eac15fbc373592c19667e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"GPT-4o","Optimization Type":"Query-dependent"}. This record does not assert independent reproduction. ### TextGrad Accuracy: 84. Result ID: 6a1eac15fbc373592c19667f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"GPT-4o","Optimization Type":"Template"}. This record does not assert independent reproduction. ### APE Accuracy: 84. Result ID: 6a1eac15fbc373592c196680. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"GPT-4o","Optimization Type":"Template"}. This record does not assert independent reproduction. ### LLM only Accuracy: 83. Result ID: 6a1eac15fbc373592c19667b. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"GPT-4o","Optimization Type":"None"}. This record does not assert independent reproduction. ### Self-Refine Accuracy: 81. Result ID: 6a1eac15fbc373592c19667d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"GPT-4o","Optimization Type":"Query-dependent"}. This record does not assert independent reproduction. ### evaluation-instructed prompt optimization Accuracy: 70. Result ID: 6a1eac15fbc373592c196670. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaMA-3","Optimization Type":"Query-dependent"}. This record does not assert independent reproduction. ### evaluation-instructed prompt optimization Accuracy: 69. Result ID: 6a1eac15fbc373592c196676. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaMA-3.1","Optimization Type":"Query-dependent"}. This record does not assert independent reproduction. ### Self-Refine Accuracy: 63. Result ID: 6a1eac15fbc373592c196671. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaMA-3","Optimization Type":"Query-dependent"}. This record does not assert independent reproduction. ### Pro-Refine Accuracy: 63. Result ID: 6a1eac15fbc373592c196672. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaMA-3","Optimization Type":"Query-dependent"}. This record does not assert independent reproduction. ### Self-Refine Accuracy: 63. Result ID: 6a1eac15fbc373592c196677. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaMA-3.1","Optimization Type":"Query-dependent"}. This record does not assert independent reproduction. ### Pro-Refine Accuracy: 63. Result ID: 6a1eac15fbc373592c196678. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaMA-3.1","Optimization Type":"Query-dependent"}. This record does not assert independent reproduction. ### APE Accuracy: 61. Result ID: 6a1eac15fbc373592c19667a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaMA-3.1","Optimization Type":"Template"}. This record does not assert independent reproduction. ### TextGrad Accuracy: 58. Result ID: 6a1eac15fbc373592c196673. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"LLaMA-3","Optimization Type":"Template"}. This record does not assert independent reproduction. ### TextGrad Accuracy: 58. Result ID: 6a1eac15fbc373592c196679. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"LLaMA-3.1","Optimization Type":"Template"}. This record does not assert independent reproduction. ### LLM only Accuracy: 56. Result ID: 6a1eac15fbc373592c196675. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"LLaMA-3.1","Optimization Type":"None"}. This record does not assert independent reproduction. ### LLM only Accuracy: 55. Result ID: 6a1eac15fbc373592c19666f. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"LLaMA-3","Optimization Type":"None"}. This record does not assert independent reproduction. ### APE Accuracy: 55. Result ID: 6a1eac15fbc373592c196674. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaMA-3","Optimization Type":"Template"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.