Benchmark Explorer

A comprehensive legal reasoning benchmark that evaluates models across six fundamental categories of legal analysis: issue-spotting, rule-recall, rule-conclusion, rule-application, interpretation, and rhetorical understanding. This crowd-sourced collection of legal reasoning tasks provides a thorough assessment of a model's ability to handle complex legal reasoning scenarios, from identifying relevant legal principles to analyzing their application and understanding legal argumentation.

Source:

Model Performance

Gemini-2.5-Pro-0325

83.6%

Gemini-2.5-Pro-Thinking

83.6%

Grok-4-Thinking

83.4%

Grok-4

83.4%

Gemini-2.5-Flash

82.8%

OpenAI-O3-high

82.5%

OpenAI-O3-medium

82.5%

Grok-3-Mini-Beta

82.0%

Grok-3-Beta

82.0%

#10

Grok-3-Think

82.0%

#11

GPT-4.1

81.9%

#12

Claude-4.1-Opus

81.9%

#13

Gemini-2.5-Flash-Thinking

81.8%

#14

Claude-4.0-Opus

81.7%

#15

GPT-5-mini

81.7%

#16

Claude-4.0-Sonnet

81.5%

#17

Claude-4.0-Sonnet-Thinking

81.3%

#18

Kimi-K2-Instruct

80.5%

#19

DeepSeek-V3

80.1%

#20

DeepSeek-V3-0324

79.9%

#21

Gemini-2.0-Flash

79.7%

#22

Claude-3.7-Sonnet-Thinking

79.3%

#23

Qwen-3

79.2%

#24

Qwen-3-Thinking

79.2%

#25

OpenAI-O4-mini-high

79.0%

#26

OpenAI-O4-mini-medium

79.0%

#27

Llama-3.1-405B

79.0%

#28

Qwen-3-Max-Preview

78.9%

#29

Claude-3.5-Sonnet-1022

78.8%

#30

Cohere-Command-A

78.7%

#31

OpenAI-O1-mini

78.7%

#32

Claude-3.7-Sonnet

78.1%

#33

Llama-3.3-70B

78.0%

#34

OpenAI-O1-1217

77.6%

#35

Claude-3-Opus

77.5%

#36

Llama-4-Maverick-17B

77.2%

#37

GPT-4-mini

76.2%

#38

Claude-3.5-Haiku

73.0%

#39

Claude-3-Sonnet

72.3%

#40

Gemini-1.5-Pro

71.6%

#41

OpenAI-O3-mini-medium

70.9%

#42

OpenAI-O3-mini-high

70.9%

#43

Cohere-Command-R-Plus

70.5%

#44

DeepSeek-R1

69.9%

#45

GPT-3.5-Turbo

64.8%

#46

Mistral-Medium-3.1

61.3%

#47

Llama-2-70B

59.4%

#48

Llama-2-13B

51.6%

#49

Llama-2-7B

49.3%