Kairos: A Native World Model Stack for Physical AI

ACE Robotics

Research official + media 3 src. ~1 min

Kairos is a full-stack world model architecture for physical AI, introducing a Cross-Embodiment Data Curriculum (open-world video → human behavior → robot interaction) and a Hybrid Linear Temporal Attention mechanism with provable error-accumulation bounds. The 4B-parameter model runs on-device in real time and tops four embodied-intelligence benchmarks including RoboTwin 2.0 (96.1%) and LIBERO-Plus.

Why it matters

712 upvotes on HuggingFace Daily — the highest among June 18 papers. First open-source world model to close the perception-to-action loop on-device without intermediate translation latency.

Importance: 4/5

Top HF Daily paper (712 upvotes) + state-of-the-art results on multiple embodied intelligence benchmarks.

Sources