RuleSafe-VL - Decision State Prediction: leaderboard

Metric: Macro-F1 (%; decidable versus underdetermined image-text cases, mean of three policy families). Source: arxiv.org. Saturation forecast: Around 2030. 10 models tracked.

Top models

#ModelScore
1Qwen 3 VL 4B64.5
2Qwen 3 VL 8B59.2
3Claude Opus 4.658.6
4GPT-5.451.7
5Gemini 3.1 Pro (Preview)38.6
6Qwen 2.5 VL 7B Instruct34.6

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