Beijing-based startup Dexmal has launched its first embodied native large model, DM0, along with the embodied native development framework Dexbotic 2.0 and the embodied native application mass production workflow, DFOL.

The company's CEO Tang Wenbin noted that 2026 will not be the dawn of embodied intelligence, but rather the dawn of embodied nativeness.

The company’s newly released DFOL (Distributed Field Online Learning) workflow is designed to enable robots to maintain high efficiency and certainty while also achieving human-like flexibility and adaptability. The core concept is the "hardware universality + model intelligence" approach.