WorldArena Track 1 - Controllability: leaderboard
WorldArena Track 1 controllability dimension averaging instruction following and semantic alignment.
Metric: Score (%). Source: huggingface.co. 124 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | UNIS | 91.6 | |
| 2 | DexWorldEngine | 91.03 | |
| 3 | ABot-PhysWorld (text) | 90.84 | |
| 4 | MAI | 90.56 | |
| 5 | RevelioWM | 90.5 | |
| 6 | PortkeySim | 90.47 | |
| 7 | Veo3.1 | 90.47 | |
| 8 | WAI | 90.44 | |
| 9 | PensieveWM | 90.4 | |
| 10 | UBWorld | 90.38 |
Interactive version: theaggregate.ai/benchmark?slug=worldarena-track-1-controllability · How It Works · Data refreshed daily, snapshot 2026-10-20.