SafeRelBench - Containment: leaderboard
Metric: Safety success rate (%) on containment-relation samples: episodes that both complete the BDDL task goal and satisfy every process-level and termination-level safety condition, over all samples; closed-loop step-level planning in Behavior-1K household scenes, base prompt, deterministic decoding; higher is better. Source: arxiv.org. Saturation forecast: Around October 2027. 7 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6 | 35 |
| 2 | GPT-5.4 | 34 |
| 3 | Qwen 2.5 VL 72B Instruct | 27 |
| 4 | Gemini 3.1 Pro (Preview) | 20 |
| 5 | Qwen 3 VL 32B Instruct | 10 |
| 6 | Qwen 3 VL 8B Instruct | 5 |
| 7 | Qwen 2.5 VL 32B Instruct | 1 |
Interactive version: theaggregate.ai/benchmark?slug=saferelbench-containment · How It Works · Data refreshed daily, snapshot 2026-09-29.