WBench - Physical Plausibility: leaderboard
The physical plausibility score of WBench (2026), a multi-turn benchmark for interactive video world models: text-driven, camera-controlled and action-conditioned video world models extend a scene turn by turn under their own controls. The score is the mean of causal fidelity and visual plausibility, on a 0-100 scale; higher is better.
Metric: Score (0-100): the mean of causal fidelity and visual plausibility, on the 158 navigation cases, multi-turn generation, each model under its own text, camera or action control; higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 20 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Wan2.7 | 71.8 | |
| 2 | LingBot-World | 71.2 | |
| 3 | Kling 3.0 | 69.3 | |
| 4 | LongCat-Video | 68.9 | |
| 5 | Seedance 1.5 | 68.3 | |
| 6 | Cosmos 2.5 | 67.4 | |
| 7 | HY-Video 1.5 | 67.3 | |
| 8 | Fantasy-World | 66.8 | |
| 9 | HY-World 1.5 | 66.3 | |
| 10 | Genie 3 | 65.7 | |
| 11 | YUME 1.5 | 65.2 | |
| 12 | InSpatio-World | 65.2 | |
| 13 | LTX-2.3 | 64.8 | |
| 14 | Happy Oyster | 63.5 | |
| 15 | HY-GameCraft | 62.4 |
Interactive version: theaggregate.ai/benchmark?slug=wbench-physical-plausibility · How It Works · Data refreshed daily, snapshot 2026-11-21.