Autonavi has recently officially launched the General World Model Workshop - ABot-World Studio. The product's biggest highlight lies in integrating interactive video generation with 3D scene construction. Users only need to input simple text or images to quickly generate an AI digital world that is highly realistic and supports real-time interaction.

The workshop has achieved a major breakthrough in its technical architecture. It not only supports local deployment on a single 5090 GPU, but also breaks the previous limitations of world models on inference duration. In official tests, the continuous inference runtime has stabilized for over one hour, finally overcoming the awkwardness of previous similar models only being able to generate "short videos."

image.png

ABot-World Studio integrates two core models, ABot-World0 and ABot-3DWorld0, which not only solve the problem of long-sequence video degradation, but also natively output 3DGS spatial assets with real geometric structures. This makes the generated world no longer just a stream of pixels, but a "physical entity" with physical rules and spatial depth, allowing users to freely explore within photo-quality scenes, just like in a game.

To increase the fun of exploration, the workshop also introduces a unique "Time-Space Portal" mechanism. Users can freely switch between different 3D scenes, connecting previously isolated scenes into an endless network through "space teleportation," such as directly linking a wooden door in a Jiangnan water town to a cyber city.

Currently, the ABot-World series of models are fully open source and have been launched on platforms such as Reactor and Hugging Face. This technological achievement is expected to show great application potential in embodied intelligence training, film storyboarding, cultural tourism, and education industries. It allows users to truly transition from mere "content viewers" to "participants" in the digital world.