RoboTrustBench - Constraint-Sensitive (GPT-5.4 Judge): leaderboard
Metric: Overall average (0-100) of the five dimension means over the same 12 criteria scored by a GPT-5.4 judge on the Constraint-Sensitive (feasible but ambiguous, occluded, cluttered or trajectory-constrained instruction) scenario, automatic evaluation that the paper runs to scale to the full benchmark (345 pairs in this scenario; the judge sees the instruction, the initial image and 20 uniformly sampled frames and cites frame evidence before scoring each criterion 1 to 5, normalized to 0-1 and printed x100 here): an image-to-video world model generates a robot-arm manipulation video from a real DROID initial frame and instruction, rated for scene entity alignment, spatiotemporal consistency, interaction rationality, task execution and visual quality; higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 7 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Kling 2.6 | 86 |
| 2 | Veo 3.1 Fast | 83.5 |
| 3 | Cosmos-Predict2.5-2B | 80.5 |
| 4 | Cosmos-Predict2.5-14B | 80.2 |
| 5 | LingBot-World | 79 |
| 6 | Wan2.2-I2V-A14B | 78.9 |
| 7 | HunyuanVideo-1.5 | 77.9 |
Interactive version: theaggregate.ai/benchmark?slug=robotrustbench-constraint-sensitive-gpt-5-4-judge · How It Works · Data refreshed daily, snapshot 2026-09-29.