Criminal Sentencing and Legal Validity Evaluation on LeCaRD v2 benchmark leaderboard · SOTA2 Research

Criminal Sentencing and Legal Validity Evaluation on LeCaRD v2

9.98RMSEL4L

Evaluation Results

Method	Links
L4L2025.11		9.98	96.2
LexGuard2026.05		9.98	96.2
L4L2025.11		12.72	94.6
LexGuard2026.05		12.72	94.6
GPT 5.22025.11		13.87	94
GPT 3.52026.05		13.87	94
GPT 5.22025.11		14.54	94.2
GPT 3.52026.05		14.54	94.2
DeepSeek v32025.11		15.42	93
DeepSeek v32026.05		15.42	93
GPT-4o2025.11		22.68	95
GPT-4o2026.05		22.68	95
Claude 4 Sonnet2025.11		26.2	94
Claude 4 Sonnet2026.05		26.2	94
Claude 4 Sonnet2025.11		26.25	93
Claude 4 Sonnet2026.05		26.25	93
DeepSeek v32025.11		26.73	64.3
DeepSeek v32026.05		26.73	64.3
GPT o4-mini2025.11		28.03	94
GPT o4-mini2026.05		28.03	94
LexiLaw2025.11		28.51	90
LexiLaw2026.05		28.51	90
GPT-4o2025.11		32.84	94
GPT-4o2026.05		32.84	94
GPT o4-mini2025.11		33.2	92
GPT o4-mini2026.05		33.2	92
DISC-LawLLM2025.11		37.14	90
DISC-LawLLM2026.05		37.14	90
LexiLaw2025.11		39.09	92.13
LexiLaw2026.05		39.09	92.13
DISC-LawLLM2025.11		52.45	83
DISC-LawLLM2026.05		52.45	83