ContrAR (Multiple Frames) - Safety Inspection: leaderboard

ContrAR (2026) tests whether vision-language models detect contradictory virtual content attacks in augmented reality, answering yes or no on real AR videos of which half carry an overlay that contradicts the physical scene, given the first, middle and last frames of each video. Detection accuracy in percent on the safety inspection videos; higher is better.

Metric: Detection accuracy (%) for contradictory virtual content attacks in AR videos (42 safety inspection videos; half the videos carry a contradiction, a balanced yes/no decision), first, middle and last frames given (multi-frame prompting); mean of three runs; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 11 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1GPT-4.188.1#505
2GPT-4o80.95#751
3Claude Sonnet 4.580.95#261
4GPT-578.57#178
5Grok 2 Vision76.19#1437
6Gemini 2.5 Pro66.67#276
7Claude Haiku 4.564.29#583
8Grok 464.29#319
9Gemini 2.5 Flash59.52#463
10Qwen 2.5 VL 7B Instruct59.52#1902
11Qwen 2.5 VL 72B Instruct54.76#841

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=contrar-multiple-frames-safety-inspection · How It Works · Data refreshed daily, snapshot 2026-11-21.