Kairos: A Regret-Aware Native World-Action Model Stack for Physical AIKairos：面向 Physical AI 的具备后悔感知能力的原生世界-动作模型栈Learning, Maintaining, and Deploying Control-Sufficient World States Kairos TeamAbstract 摘要World models are becoming a central substrate for Physical AI, yet their current development is differentiating across representational, generative, interactive, and unified understandinggenerationprediction world models.While these paradigms have advanced abstract world reasoning, high-fidelity visual generation, interactive simulation, and embodied world-action modeling, none alone provides the deployable, long-horizon, control-sufficient state required by embodied agents. We intr