SafeSceneReason: leaderboard

Metric: Accuracy (%; micro average over all 1,000 questions; single-answer multiple choice on the 1,000-question test set balanced across answer positions, built from workplace scene graphs and NIOSH FACE and CSB accident-report figures; temperature 0, unparseable answers counted wrong). Source: arxiv.org. Saturation forecast: Around December 2026. 8 models tracked.

Top models

#ModelScore
1GPT-5.6 Sol88.7
2GPT-5.587.9
3Gemini 3.1 Pro (Preview)80.3
4Gemini 3.5 Flash78.7
5Claude Fable 574.1
6Kimi K2.670.7
7Qwen 3.6 27B40.9
8Qwen 3.5 9B37.4

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