ResearchBenchmarksReasoning on Big-Bench Hard (Subtask Mean/Std Reporting)Follow100BBH: Boolean Expressions (Mean)PSAO.LLM98.200898.667999.13599.6021May 14, 2026Evaluation ResultsMethodMethodLinksBBH: Boolean Expressions (Mean)BBH: Boolean Expressions (Std)BBH: Causal Judgement (Mean)BBH: Causal Judgement (Std)PSAO.LLMModel=Gemini-2-FlashModel=Gemini-2-Flash2026.05100072.731.82COPROModel=Gemini-2-FlashModel=Gemini-2-Flash2026.0598.41.0568.913.58MIPROModel=Gemini-2-FlashModel=Gemini-2-Flash2026.0598.271.2667.093.37