ContrAR (Single Frame) - Indoor Navigation: leaderboard
Metric: Detection accuracy (%) for contradictory virtual content attacks in AR videos (54 indoor navigation videos; half the videos carry a contradiction, a balanced yes/no decision), one middle frame of each video given (single-frame prompting); mean of three runs; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Flash | 85.19 |
| 2 | GPT-4o | 83.33 |
| 3 | GPT-5 | 81.48 |
| 4 | GPT-4.1 | 79.63 |
| 5 | Gemini 2.5 Pro | 75.93 |
| 6 | Grok 4 | 75.93 |
| 7 | Claude Sonnet 4.5 | 59.26 |
| 8 | Qwen 2.5 VL 72B Instruct | 55.56 |
| 9 | Qwen 2.5 VL 7B Instruct | 53.7 |
| 10 | Claude Haiku 4.5 | 50 |
Interactive version: theaggregate.ai/benchmark?slug=contrar-single-frame-indoor-navigation · How It Works · Data refreshed daily, snapshot 2026-10-07.