Skip to main content

Benchmark profile

Vals LegalBench (LegalBench)

Vals AI legal benchmark with issue, rule, conclusion, interpretation, and rhetoric task views.

Data verified

How BenchLM shows LegalBench

BenchLM mirrors the public Vals AI LegalBench leaderboard captured from https://www.vals.ai/benchmarks/legal_bench and updated by Vals on July 17, 2026. The snapshot preserves overall scores, uncertainty, latency, cost-per-test metadata, and task-level scores where Vals publishes them.

LegalBench is display only on BenchLM. Vals proprietary or Vals-hosted aggregate views are useful context, but BenchLM does not use them as weighted ranking inputs or as a replacement for benchmark-native source records.

126 Vals rows6 task viewspublic datasetTasks: Overall, Issue Tasks, Rule Tasks, Conclusion Tasks, Interpretation TasksDisplay only

LegalBench score on LegalBench — July 17, 2026

BenchLM mirrors the published legalbench score view for LegalBench. Claude Fable 5 leads the public snapshot at 88.56% , followed by Gemini 3.1 Pro Preview (87.40%) and Gemini 3 Pro Preview (87.03%). BenchLM does not use these results to rank models overall.

126 modelsExternal benchmark mirrorsCurrentDisplay onlyUpdated July 17, 2026

LegalBench score table (126 models)

Score
1
88.56%
4
86.97%
6
GPT-5.5OpenAI
86.52%
7
GPT-5.4OpenAI
86.04%
8
GPT-5OpenAI
86.02%
9
Kimi K3Moonshot AI
86.02%
10
Grok 4.5SpaceXAI
85.97%
11
GPT-5.1OpenAI
85.68%
12
MiniMax M3MiniMax
85.42%
13
85.30%
14
85.25%
15
85.11%
16
85.10%
17
84.98%
18
84.91%
19
Kimi K2.6Moonshot AI
84.74%
21
Grok 4.3SpaceXAI
84.46%
22
GLM 5.1Zhipu AI
84.39%
24
84.28%
25
84.23%
26
84.22%
28
GLM 5.2Zhipu AI
84.07%
29
84.06%
30
84.03%
31
83.98%
32
83.92%
35
O3OpenAI
83.76%
36
83.60%
37
83.57%
38
83.46%
39
GLM 4.7Zhipu AI
83.36%
40
Grok 4 0709SpaceXAI
83.19%
42
GPT-4.1OpenAI
83.10%
43
Claude Opus 4Anthropic
83.07%
44
82.95%
45
InklingThinkingmachines
82.95%
46
82.84%
47
GPT-5.2OpenAI
82.76%
49
Grok 3SpaceXAI
82.59%
52
82.45%
53
GPT-4oOpenAI
82.21%
54
82.12%
59
Qwen3 MaxAlibaba
81.86%
60
81.77%
61
81.45%
63
DeepSeek V3Fireworks AI
80.76%
64
80.60%
65
80.46%
66
O1OpenAI
80.39%
67
80.33%
68
80.32%
69
Kimi K2 ThinkingMoonshot AI
80.20%
70
Qwen3 235b A22bFireworks AI
80.18%
71
GPT-4oOpenAI
80.12%
72
80.00%
73
79.96%
74
79.70%
75
GLM 4.6Zhipu AI
79.61%
77
O4 MiniOpenAI
79.19%
78
79.14%
81
Mimo V2.5Xiaomi
78.89%
83
78.36%
84
78.04%
85
77.92%
86
77.81%
87
77.74%
88
DeepSeek V3 0324Fireworks AI
77.73%
90
77.13%
91
76.08%
92
GPT Oss 120bFireworks AI
75.94%
93
GLM 4.5Zhipu AI
75.63%
94
75.45%
95
Laguna M.1Poolside
75.14%
96
74.16%
98
71.96%
99
O3 MiniOpenAI
71.54%
100
Laguna Xs.2Poolside
71.03%
101
GPT Oss 20bFireworks AI
70.85%
102
70.33%
103
70.25%
104
69.73%
105
69.56%
106
69.16%
107
69.08%
108
68.29%
109
68.14%
110
67.98%
111
DeepSeek R1Fireworks AI
67.32%
112
66.62%
113
DeepSeek V3p2Fireworks AI
65.96%
114
64.37%
115
63.24%
116
62.43%
117
61.98%
118
61.06%
119
55.84%
120
54.77%
121
53.77%
122
53.70%
123
51.87%
124
50.13%
125
40.02%
126
Command RCohere
32.97%

The published LegalBench snapshot places Claude Fable 5 first at 88.56%. The third row is 1.54 points behind. The broader top-10 range is 2.60 points, so many of the published results sit in a relatively narrow band.

126 models have been evaluated on LegalBench. The benchmark falls in the External benchmark mirrors category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. LegalBench is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About LegalBench

Year

2026

Tasks

Legal reasoning task views

Format

Accuracy score

Difficulty

Professional legal reasoning

BenchLM mirrors Vals LegalBench as display-only legal-domain evidence and does not use it in weighted rankings.

BenchLM freshness & provenance

Version

LegalBench 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

FAQ

What does LegalBench measure?

Vals AI legal benchmark with issue, rule, conclusion, interpretation, and rhetoric task views.

Which model leads the published LegalBench snapshot?

Claude Fable 5 currently leads the published LegalBench snapshot with 88.56% legalbench score. BenchLM shows this benchmark for display only and does not use it in overall rankings.

How many models are evaluated on LegalBench?

126 AI models are included in BenchLM's mirrored LegalBench snapshot, based on the public leaderboard captured on July 17, 2026.

Last updated: July 17, 2026 · mirrored from the public benchmark leaderboard

Choose a model with this week’s evidence

Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.

One email each week. Unsubscribe anytime.