SpatialTrust: leaderboard

Metric: Overall score (%; mean over the five question types of sensitive-factor detection, direct and indirect factor identification and explanation; over SpatialTrust's 577 secure-authentication images (five questions each); IoU threshold 0.5 for grounded evidence). Source: arxiv.org. Saturation forecast: Around 2031. 8 models tracked.

Top models

#ModelScore
1Qwen 3 VL 30B A3B Instruct36.78
2Gemini 2.5 Flash33.81
3Qwen 3 VL 8B Instruct28.91
4GPT-5.425.94
5GLM-4.1V-9B (Thinking)21.62
6Qwen 2 VL 7B Instruct15.89

Interactive version: theaggregate.ai/benchmark?slug=spatialtrust · How It Works · Data refreshed daily, snapshot 2026-09-26.