On July 9, Lingbo Technology, a subsidiary of Ant Group, unveiled the open-sourcing of two groundbreaking models: LingBot-World 2.0 and LingBot-Video. LingBot-World 2.0 represents a leap forward in real-time interactive world simulation, enabling hour-long dynamic scene generation with 720p/60fps high-definition output. Notably, it pioneers the integration of an Agent mechanism into world simulation models, enhancing adaptability and user interaction. Meanwhile, LingBot-Video emerges as the world's first open-source video synthesis foundational model built on the Mixture-of-Experts (MoE) architecture. Specifically designed for embodied AI applications, it provides a robust open-source framework for robots to comprehend actions, tasks, and the physical environment, marking a significant advancement in AI-driven video generation.
