NTU & CUHK Release Apple-π to Test if Video Models Truly Understand Physics
机器之心 · wechat · 2026-07-30
A joint team from NTU and CUHK released Apple-π, a benchmark designed to evaluate whether video generation models genuinely understand physical laws rather than merely mimicking visual statistical regularities.
The benchmark includes 400 classical mechanics cases and breaks down scientific reasoning into three stages: perception, law formulation, and deduction. Tests on 11 models reveal a clear "reasoning funnel" where perception is easiest and deduction is hardest. The best video model scored only 0.473, and even top unified models struggled with dynamic deduction, highlighting a significant gap before video models can become reliable world simulators.
More from Multimodal
- Chinese AI Model Shows Off 'The Greatest Free Kick' — ZabihullahAtal · 2026-07-30
- Sakana AI and NYU Release Dream-Cubed: AI Generates Playable Minecraft Worlds Trained on Billions of Cubes — tkasasagi · 2026-07-30
- AI-Generated Short Film: A Japanese Shopping Street That Never Existed — writingdeveloper · 2026-07-30
- AI-Generated Anime Short Clip: Realmz of the Redeemers — Kindly_Poet_7878 · 2026-07-30
- NVIDIA's PDD Paper: Parallel Decoding Distillation for Faster Image/Video Generation — Scriabinical · 2026-07-30
- Royal Family AI Slop Microdramas Are Weirdly Addictive — venturetwins · 2026-07-30