15 AI models tested on moral dilemmas

Each model answered the same dilemmas. Open a profile to see its choices, reasoning, and where it stands out.

• Qwen 3.7 MaxAlibaba (Qwen)
• Claude 4.8 OpusAnthropic
• Claude 4.6 SonnetAnthropic
• DeepSeek v4 ProDeepSeek
• Gemini 3.5 FlashGoogle DeepMind
• Gemini 3.1 Pro PreviewGoogle DeepMind
• GPT 5.5OpenAI
• GPT 4oOpenAI
• GPT o3OpenAI
• Grok 4.3xAI
• GLM 5.2Zhipu AI (Z.ai)
• MiniMax M3MiniMax
• Llama 3.3 70BMeta
• Hermes 4 405BNous Research
• Nemotron 3 Ultra 550BNVIDIA

Data as of 2026-06-18. Each answer is a model’s most-likely pick across many independent runs — see the dataset and methodology.