RoboJailBench: leaderboard
Harmonic mean (%) of the security rate (adversarial goals rejected) and the utility rate (benign goals accepted) in RoboJailBench (2026), where each robot-scene image from six embodied datasets comes with a benign and an adversarial instruction and the vision-language model must answer okay or deny. Higher is better.
Metric: Security-utility harmonic mean (%) of the rejection rate on adversarial goals and the acceptance rate on benign goals, over the paired benign and adversarial goals of six robot-scene datasets (RoboVQA, DROID, NVIDIA PhysicalAI AV, RH20T and Robo2VLM augmented with intent-contrast prompts, and the new RJB-Instructions), each image with one benign and one adversarial instruction and a forced okay or deny answer, no attack and no defense; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 5 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | GPT-5.4 Nano (2026-03-17) | 97.84 | #473 | |
| 2 | Claude Haiku 4.5 | 96.57 | #583 | |
| 3 | GPT-5.4 Mini | 95.93 | #406 | |
| 4 | Gemini 3 Flash (Preview) | Non-reasoning | 95.37 | #152 |
| 5 | gemini-robotics-er-1.6-preview | Non-reasoning | 92.75 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=robojailbench · How It Works · Data refreshed daily, snapshot 2026-11-21.