RoboAbstention - Contradictory Instructions: leaderboard
The Contradictory Instructions category of RoboAbstention (2026, The Yes-Man Syndrome): 837 instructions that contradict themselves, given with an image to a vision-language robot planner that should abstain or ask. The score is the abstention rate in percent; higher is better.
Metric: Abstention rate (%) on the 837 Contradictory Instructions instructions, image plus instruction to an embodied VLM planner, every instruction warranting abstention, act or abstain judged by GPT-5.4 mini (97.5 percent agreement with human labels), API default sampling; higher is better. Source: arxiv.org. 8 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Gemini 2.5 Flash | 36.3 | #466 | |
| 2 | GPT-5.4 | 29 | #138 | |
| 3 | Llama 4 Maverick | 28.6 | #1113 | |
| 4 | Claude Sonnet 4.6 | 26.5 | #179 | |
| 5 | GPT-5.4 Nano | 22.6 | #653 | |
| 6 | GPT-5.4 Mini | Low | 22 | #409 |
| 7 | Gemini Robotics-ER 1.6 | 16.6 | #175 | |
| 8 | Qwen 3.5 27B | 5.4 | #380 |
Interactive version: theaggregate.ai/benchmark?slug=roboabstention-contradictory-instructions · How It Works · Data refreshed daily, snapshot 2026-11-22.