ResearchBenchmarksScience on StemEZ physical chemistry MMLU-ProFollow86.57ScoreAgentSPEX82.72283.72184.7285.719Apr 14, 2026Evaluation ResultsMethodMethodLinksScoreAgentSPEXModel=GPT-5, Domain=Sc...Model=GPT-5, Domain=Science2026.0486.57ReActModel=GPT-5, Domain=Sc...Model=GPT-5, Domain=Science2026.0484.72CoTModel=GPT-5, Domain=Sc...Model=GPT-5, Domain=Science2026.0482.87