SafeSceneReason - Comparative Reasoning: leaderboard
Metric: Accuracy (%; the 108 comparative reasoning 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
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 | 86.11 |
| 2 | GPT-5.6 Sol | 80.56 |
| 3 | Gemini 3.5 Flash | 78.7 |
| 4 | Gemini 3.1 Pro (Preview) | 76.85 |
| 5 | Kimi K2.6 | 71.3 |
| 6 | Claude Fable 5 | 67.59 |
| 7 | Qwen 3.6 27B | 26.85 |
| 8 | Qwen 3.5 9B | 25 |
Interactive version: theaggregate.ai/benchmark?slug=safescenereason-comparative-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-26.