Robbyant, an embodied AI company within Ant Group, has open-sourced LingBot-World 2.0 (Infinity), the latest version of its interactive world model, adding support for hour-long real-time world generation, high-definition output and expanded interactive capabilities.
The company says the new model represents a significant advance over LingBot-World 1.0, extending stable world generation from minutes to continuous hour-long sessions while maintaining visual quality and supporting real-time user interaction.
According to Robbyant, LingBot-World 2.0 produces 720p video at 60 frames per second and is designed to generate, stream and display content simultaneously, reducing latency and allowing users to interact with the environment as it is created.
The model is available through the company’s Reactor platform, where users can control character movement and change viewpoints in real time using a keyboard.
Robbyant says LingBot-World 2.0 introduces a native agent architecture intended to make generated environments continuously interactive rather than simply viewable.
The system includes a dual-agent mechanism comprising a Pilot Agent, which plans and executes character actions, and a Director Agent, which dynamically introduces new events as the virtual world evolves.
The expanded action set includes activities such as attacking, jumping, gliding, casting spells and shooting arrows, while users can also trigger events including weather changes, day-night cycles and the introduction of new entities through text prompts.
The company says the model is built on a causal pretraining approach combined with its proprietary Mask of Bidirectional Attention (MoBA) mechanism, which is designed to reduce the visual degradation that typically occurs during long-duration AI world generation.
Robbyant says internal stress testing demonstrated stable visual fidelity throughout hour-long generation sessions without noticeable quality drift.
The release also supports multiple users sharing a persistent virtual environment, enabling collaborative interaction within the same AI-generated world.
Alongside LingBot-World 2.0, Robbyant has also open-sourced LingBot-Video, which it describes as the first open-source video generation foundation model based on a Mixture-of-Experts architecture developed specifically for embodied AI.
According to the company, LingBot-Video has been designed for robotics applications, improving inference efficiency, physical realism, action understanding and task completion as AI-generated video models increasingly move beyond content creation toward real-world robotic systems.
