Locally Runnable World Model Released

dr_cintas · x · 2026-07-10

Robbyant has released LingBot-World 2.0, an action-conditioned world model. By inputting a single image, a stream of camera poses, and text prompts, the model autoregressively generates an explorable world.

The release features 14B open-source weights on Hugging Face, associated agentic harness code, and a lightweight 1.3B version optimized for a single consumer-grade GPU. The author also highlights that it showed almost no noticeable degradation during a 60-minute continuous run test.

Related event: Ant-backed Team Open-Sources LingBot-World 2.0 Real-time World Model(26 posts)→

Original post →

More from coding & agent

coding & agent channel →