Legal Reasoning
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
74.9Memoization
35
Apr 21, 2026
89.22Test Accuracy
22
Mar 10, 2026
90Accuracy
18
Jun 2, 2026
83.13Accuracy (CaseHold)
16
May 12, 2026
0.033MAE
16
May 11, 2026
86.46Accuracy
16
Apr 9, 2026
79.3Balanced Accuracy
16
Jun 1, 2026
65.42Accuracy
14
Mar 19, 2026
34.4LLM-as-judge Score
13
Jun 2, 2026
26.52Score
13
Mar 30, 2026
82.1Accuracy
12
Mar 23, 2026
57.1Pass@1
10
Apr 14, 2026
23.4Pass@1 Accuracy
10
Apr 14, 2026
75.5Accuracy
10
Feb 26, 2026
70Micro-F1
10
Feb 26, 2026
27.5Accuracy
9
Jun 30, 2026
54.4Accuracy+
9
Jun 30, 2026
69Exact Match
8
May 26, 2026
96Cumulative Score (CS)
8
Mar 10, 2026
82.7Accuracy (Cardl)
6
Feb 26, 2026
85LegalBench Score
5
Jun 4, 2026
90Accuracy
5
Apr 28, 2026
92Accuracy
5
Apr 28, 2026
45.29Score
5
Mar 30, 2026
0.377GPT-5 Grader Score
4
Jun 25, 2026