List your company

Multiple Choice Question Answering on LegalMC4 (test)

72Exact AccuracyGPT-5 (min. reasoning)
59.41662.68365.9569.217Jan 20, 2026

Evaluation results

MethodLinks
GPT-5 (min. reasoning)Train Data=Base Model2026.0172
Gemma 3 (12B)Train Data=Difficulty-Graded QA Data2026.0171.2
GPT-5-mini (min. reasoning)Train Data=Base Model2026.0170.9
Gemma 3 (12B)Train Data=Standard Instruction QA Data2026.0166.9
Gemma 3 (12B)Train Data=Base Model2026.0166.8
LLaMA 3.1 (8B)Train Data=Difficulty-Graded QA Data2026.0166.1
LLaMA 3.1 (8B)Train Data=Base Model2026.0160.9
LLaMA 3.1 (8B)Train Data=Standard Instruction QA Data2026.0159.9