SpatialTrust - Indirect Factor Explanation: leaderboard

Metric: Grounded explanation score (%; correctly explaining why an indirect factor is sensitive; over SpatialTrust's 577 secure-authentication images (five questions each); IoU threshold 0.5 for grounded evidence). Source: arxiv.org. Saturation forecast: Around 2033. 8 models tracked.

Top models

#ModelScore
1Qwen 3 VL 30B A3B Instruct22.91
2Qwen 3 VL 8B Instruct10.92
3Gemini 2.5 Flash10.6
4GLM-4.1V-9B (Thinking)6.21
5GPT-5.44.53
6Qwen 2 VL 7B Instruct0.87

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