← BackThe DecoderJonathan KemperOct 18, 2026, 16:49 UTC+8AI agents have bad taste in long-term decisions and a new benchmark proves itTaste-Bench 用 502 道二选一问题评测 AI 智能体的长期决策判断能力,14 个受测模型中表现最好的 GPT-5.6 Sol 准确率为 59.7%。Read the original ↗ShareCopy linkShare imageExport MarkdownSee every follow-up