LingBot-Video Predicts Robot Actions
Extra-Avocado8967 · reddit · 2026-07-13
Robbyant's open-weights LingBot-Video rolls out predicted future frames based on a given first frame and control signals.
The post focuses on its ability to map "action signals + initial image" to future frames, using it to discuss an age-old question: is this a world model, or just a video generator? The author believes it touches the core debate of world models, though there are boundary issues regarding whether it reaches the level of world models like Dreamer or JEPA.
Related event: Embodied Video Model LingBot-Video Goes Open Source(4 posts)→
More from Embodied
- WanGP v18.81 adds Tencent's Hy-Nomas world model, playable on 8GB VRAM — cocktailpeanut · 2026-10-27
- NVIDIA open-sources FLORA, a robotics data engine ~125x cheaper than GPT-6 Astra — nvidia · 2026-10-27
- Embodied AI startups need a different operational discipline, investor says — ___Mufasaa · 2026-10-27
- π0.5 study: raising demos from 10 to 438 lifts Place Ball in Bowl success from 7% to 83% — stingning · 2026-10-27
- GPT-6 Astra controls dexterous hands zero-shot at just 5.2% success in 500 trials — stingning · 2026-10-27
- HandsOn benchmark: 66 verified tasks and 18,000 human demos for dexterous manipulation — stingning · 2026-10-27