MoRa, the Agentic-Native embodied architecture developed by Mochi, has received over 1 billion in financing

MoQi Intelligence recently released the embodied model architecture MoRA, proposing an Agentic-Native technical approach to reconfigure the dual-system boundary. This architecture reduces some of the capabilities of agents to the strategy model (System 1), allowing it to directly manage targets, maintain execution memory, and assess task progress. Only when encountering capability limits or exceptions does it request intervention from the higher-level system (System 2). Through “target-condition execution,” multi-granularity memory, and a closed-loop design with progress awareness, MoRA aims to address the challenges of state maintenance and continuous execution in long-term tasks. Meanwhile, MoQi Intelligence has completed over 1 billion yuan in angel round financing, with institutions such as Alibaba and Tencent participating. The company plans to publicly demonstrate the wheeled robot MORPHI KINO equipped with MoRA at the World Robot Conference, demonstrating various continuous operations including a 15-minute household task.