Ant Lingbo open-sources real-time interactive world model LingBot-World 2.0
Original title:无限世界、实时交互,蚂蚁灵波开源 LingBot-World 2.0 世界模型
AI summary
Ant Lingbo open-sourced the new-generation real-time interactive world model LingBot-World 2.0 (14B parameters), supporting rich character actions such as casting spells, attacking, and jumping, as well as text-driven events (such as switching scenes or summoning storms). It includes a built-in Pilot Agent and Director Agent to enable continuous world evolution and supports simultaneous multi-user interaction. The model uses a causal pretraining paradigm and a hybrid bidirectional autoregressive attention mask (MoBA), stably outputting 720p/60fps real-time visuals with no quality degradation over tests lasting up to one hour. It reduces sampling costs through consistency distillation and DMD, and achieves low-latency interaction through attention kernel optimization, hybrid parallel inference, dynamic KV cache scheduling, and asynchronous streaming transmission. Model weights and inference code are open-sourced under a non-commercial license, SGLang has been adapted, and online experiences are available through Reactor PC and the Lingguang APP.
Key Takeaways
- Ant Lingbo open-sourced the new-generation real-time interactive world model LingBot-World 2.
- 0 (14B parameters),supporting rich character actions such as casting spells,attacking,and…
- It includes a built-in Pilot Agent and Director Agent to enable continuous world evolution and…
- The model uses a causal pretraining paradigm and a hybrid bidirectional autoregressive attention…
- It reduces sampling costs through consistency distillation and DMD,and achieves low-latency…
Quick Facts
- Published
- Updated
- Source
- IT之家·人工智能
- Category
- Model Releases
- AI score
- 73
蚂蚁灵波这个开源世界模型把实时世界生成推到了小时级还不崩,720p/60fps实时交互,做开放世界游戏或自动驾驶仿真的团队值得上手测。
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: Ant Lingbo open-sources real-time interactive world model LingBot-World 2.0. July 9, 2026. https://new.nat6.net/en/news/cmrd6bmv300xxih4byg33dbws/