List your company

Legal Reasoning on BarExam MBE (test)

82.1AccuracyReasoning Only
41.12451.76262.473.038Mar 20, 2026

Evaluation results

MethodLinks
Reasoning OnlyBackbone=Llama-3.3-70B2026.0382.113.949.550.41.730.8661.5
Reasoning OnlyBackbone=GPT-oss-120B2026.0377.817.733.39.431.617.942.9
Reasoning OnlyBackbone=Gemma-3-27B2026.0365.84.520.228.27.745.37.742.9
Reasoning OnlyBackbone=Gemma-3-12B2026.0360.74.22122.25.143.61236.6
EvidenceRL with GRPOBackbone=Llama-3.1-8B2026.0360.718.951.941626.54.367.6
EvidenceRL with GRPOBackbone=Gemma-3-12B2026.0360.71133.637.66.823.19.462
Reasoning OnlyBackbone=Llama-3.1-8B2026.0357.3-1.73.318.810.351.312.832.8
Reasoning OnlyBackbone=GPT-oss-20B2026.0354.7-8.3-1.515.412.853.89.428.1
EvidenceRL with GRPOBackbone=Llama-3.2-3B2026.0350.47.729.326.510.336.8655.9
EvidenceRL with GRPOBackbone=Gemma-3-4B2026.0348.77.833.826.5639.36.854.4
Reasoning OnlyBackbone=Gemma-3-4B2026.0347.95.314.519.710.344.46.841.1
Reasoning OnlyBackbone=Llama-3.2-3B2026.0342.7-3.85.313.715.446.29.432