RuleSafe-VL - Decision Relation Identification: leaderboard

Metric: Macro-F1 (%; typed relations among a policy's atomic rules from the policy text, mean of three policy families). Source: arxiv.org. Saturation forecast: Around March 2027. 11 models tracked.

Top models

#ModelScore
1Gemini 3.1 Pro (Preview)64.8
2Claude Opus 4.663.4
3GLM-5.155.8
4MiniMax-M2.750.7
5Qwen 3 VL 8B37.1
6Qwen 3 VL 4B35.2
7Qwen 2.5 VL 7B Instruct25.6

Interactive version: theaggregate.ai/benchmark?slug=rulesafe-vl-decision-relation-identification · How It Works · Data refreshed daily, snapshot 2026-09-25.