Meituan LongCat Proposes WBench: First Multi-Turn Benchmark for Interactive Video World Models
Original title:从月球漫步到赛博都市,WBench测出了世界模型的边界
AI summary
The Meituan LongCat team proposed WBench, the first systematic multi-turn evaluation benchmark for interactive video world models, including 289 test cases and 1,058 interaction rounds, covering four interaction modes: navigation, subject action, event editing and viewpoint switching.
Key Takeaways
- The Meituan LongCat team proposed WBench,the first systematic multi-turn evaluation benchmark for…
- including 289 test cases and 1
- 058 interaction rounds,covering four interaction modes: navigation,subject action,event editing…
Quick Facts
- Published
- Updated
- Source
- 公众号:龙猫LongCat(美团)
- Category
- Papers & Research
- AI score
- 65
将导航能力独立归因于空间状态表示,揭示连续交互位姿误差累积是迭代生成范式的结构性缺陷,为评估与改进交互式世界模型提供了新框架。
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: Meituan LongCat Proposes WBench: First Multi-Turn Benchmark for Interactive Video World Models. June 11, 2026. https://new.nat6.net/en/news/cmsdu2t2b011profzy9njjxyb/