MORPHI debuts at WRC with MoRA, its embodied intelligence model architecture

MORPHI (墨奇智能), a Chinese embodied intelligence company, made its first systematic domestic debut at the World Robot Conference (WRC), publicly demonstrating MoRA, its newly released embodied intelligence model architecture, according to a QbitAI report on August 20.

At the show, a robot powered by the MoRA "embodied brain" performed long-horizon tasks in a live, real-world setting — a practical test of how the architecture holds up outside the lab and MORPHI's first full public presentation of its technology in China.

MoRA is positioned as an embodied intelligence model architecture and the centerpiece of the exhibition: it aims to support robots handling complex tasks in unstructured environments through a unified model framework rather than task-specific rules.

Long-horizon task execution is one of the hardest problems in embodied AI: robots must stay stable and generalize across long, multi-step operations, and the value of the WRC live demonstration is putting the model's capability to the test in real scenes.

For MORPHI, the appearance marks a shift from technology release toward broader industry and public engagement, as it begins communicating its roadmap to the domestic market.

Embodied intelligence has become a focus of China's AI race: from model architecture and data to robot hardware, players are competing to own the "embodied brain," and MoRA adds a new variable to the field.

The questions ahead are how MORPHI opens up MoRA, which robot makers it partners with, and whether the long-horizon capability can be reproduced across more real-world scenarios.