LingBot-World
Robbyant (Ant Group) · Released January 2026
LingBot-World is Robbyant's open world simulator built on a video generation model. You give it an image and steer with camera poses or action strings; its distilled Fast variant answers in under a second, and the weights are Apache-2.0.
- Weights
- Open weightsApache-2.0
- Commercial use
- Permissive license
- Real-time
- Real-time, per authorsNot independently measured
- Where to try
- 1 placeReactor
01 Controls
What you control
- A starting image
- Camera poses
- Action strings (Act preview)
02 License
Can you use it commercially?
The weights' model card declares Apache-2.0, a permissive license that allows commercial use. Check what the model was built on, because a base model's terms can still apply.
Declared license: Apache-2.0 · Model card
A reading of the published license text, not legal advice.
03 Speed
Does it run in real time?
Real-time, per authors. Under 1 s latency while producing 16 fps at 480p or 720p.
Figures are the authors' own, with their settings and hardware. They are not comparable across models.
04 Setup
How to run it
- 01Clone robbyant/lingbot-world and install requirements.txt (PyTorch 2.4 or newer) and flash-attn.
- 02Download robbyant/lingbot-world-base-cam from Hugging Face or ModelScope.
- 03For the real-time version, also download robbyant/lingbot-world-fast and run bash run_fast.sh with the weights folder and a frame count.
- 04A quantized model is provided for machines with less GPU memory.
Summarised from the project's own instructions. Follow the source for exact commands and versions.
05 Access
Where to try it
Listed alphabetically. Windflow maintains this directory and is one of the hosts; a listing does not promise capacity at any given moment.
06 Questions
LingBot-World FAQ
Is LingBot-World open source?
LingBot-World publishes downloadable weights under Apache-2.0. The declared license is permissive.
Can I use LingBot-World commercially?
The weights' model card declares Apache-2.0, a permissive license that allows commercial use. Check what the model was built on, because a base model's terms can still apply.
Does LingBot-World run in real time?
Yes, according to its authors. Under 1 s latency while producing 16 fps at 480p or 720p. This directory has not measured it independently.
Where can I try LingBot-World?
You can try it on Reactor.
What does LingBot-World take as input?
You give it a starting image, camera poses and action strings (Act preview).
How do I run LingBot-World?
Following the official instructions: Clone robbyant/lingbot-world and install requirements.txt (PyTorch 2.4 or newer) and flash-attn. Download robbyant/lingbot-world-base-cam from Hugging Face or ModelScope. For the real-time version, also download robbyant/lingbot-world-fast and run bash run_fast.sh with the weights folder and a frame count. A quantized model is provided for machines with less GPU memory.
What are alternatives to LingBot-World?
Similar models in this directory include ABot-World (open weights, permissive license), AlayaWorld (open weights, custom terms) and DreamX-World (open weights, permissive license).
Similar world models
All 56 in the directoryExplorable world
ABot-World
AMAP CV Lab, Alibaba
Image and action-conditioned generation
Permissive licenseOpen weightsReal-time claimTry on fal, Windflow
Explorable world
AlayaWorld
Alaya Lab
Interactive world generation
Custom termsOpen weightsReal-time claimExplorable world
DreamX-World
AMAP, Alibaba
Camera control over long autoregressive rollouts
Permissive licenseOpen weightsTry on Windflow
Explorable world
Echo-Memory
JD Joy Future Academy
Starting image, prompt and camera actions
Permissive licenseOpen weightsTry on Hugging Face Space