A photo, a smartphone, and you can walk through the AI-created world.

On April 27th, the Ant Lingguang App officially launched the "Experience World Model" feature, becoming the first AGI product in the industry to experience a world model on mobile devices. Users need only upload a picture, and they can explore a 3D world for up to 60 seconds on their phone. They can also control the perspective with a mobile game, walking and exploring as if playing a game. From triggering the command to starting the exploration, it only takes a few seconds. This is the first time the industry has realized running a world model on the edge, with minute-level long-term consistency and real-time interactive experience, once again leading the AGI products. Lingguang has always been committed to exploring the boundaries of intelligence. Before this, Lingguang launched the "Flash Application" function, which was the first to realize generating an application in 30 seconds on a mobile device, leading the popular "Wish Coding".

image.png

(Figure caption: Open the Lingguang App on your phone, click the "+" at the bottom left to upload a picture, then click "Generate the World in the Picture" to experience the world model.)

A world model is considered one of the important paths toward AGI (General Artificial Intelligence) and an important bridge connecting the digital world and the physical world. The "Experience World Model" function of the Lingguang App is powered by the Ant Lingbo LingBot-World-Fast world model, which has also been open-sourced.

The Lingguang App provides users with an easy entry point for experiencing the world model. Open the Lingguang App, upload a picture in the chat box, and the system will intelligently recommend operation instructions. Users can choose "Generate the World in the Picture." Alternatively, directly input natural language such as "Help me explore this world from a first-person perspective," and the system will automatically enter the world model generation process. From triggering the command to starting the exploration, it only takes a few seconds.

After entering the world model experience page, the Lingguang App has carefully designed operations for mobile users, introducing an innovative mobile game joystick control method, allowing users to explore the AI-generated 3D world in the most familiar way. Specifically, the joystick on the left side of the screen controls the character's movement within the 3D scene, allowing users to freely walk forward, backward, left, and right; the joystick on the right side controls the rotation of the view, enabling comprehensive exploration. This control logic is highly consistent with mainstream 3D mobile games, allowing players to operate without additional learning, achieving "zero threshold immersion."

Deploying world models on mobile devices is a recognized challenge in the industry. High computing power requirements, difficult latency control, and varying terminal performance are all practical engineering challenges. The Lingguang team used efficient and low-latency streaming transmission technology to achieve a response delay of less than 100 milliseconds, allowing users to start exploring the 3D world within seconds after triggering the command, breaking the previous stereotype that world models were "high barrier, high computing power, and hard to implement."

Cai Wei, the head of the Lingguang App, said, "The 'Experience World Model' feature is another practice of Lingguang in exploring the boundaries of intelligence. Previously, the 'Flash Application' function of Lingguang could generate an application in natural language within 30 seconds, which also gave ordinary users the coding capabilities that were originally limited to professional developers. Lingguang hopes to continuously explore the boundaries of intelligence, uncover unmet user needs, and provide good AI experiences to everyone."

Currently, users can download the Lingguang App from major app stores and directly experience the world model feature.