ResearchBenchmarksToxicity Evaluation on RealToxicityPrompts (Nontoxic Subset)Follow0.2Toxic FractionGPT-4o-m0.140.5450.951.355Jul 15, 2024Evaluation ResultsMethodMethodLinksToxic FractionGPT-4o-m2024.070.2GPT-4o2024.070.3Gem-Fl-Lt2024.070.5Gem-Fl2024.071.1Gem-Pro2024.071.7