SafePyramid - L1 Rule Dependencies: leaderboard
Metric: Rule matching rate (%) on the 1,000 L1 cases (policies that add exception and conditional rules): per case (a multi-turn conversation plus an in-context safety policy), the predicted set of violated rules matches the ground-truth set when neither false positives nor false negatives exceed (1 - tau) of the true violations; averaged over tau = 0.7, 0.8, 0.9 and 1.0; per-policy evaluation, refused cases excluded; higher is better. Source: arxiv.org. Saturation forecast: Around April 2027. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 (xHigh) | 56.8 |
| 2 | Hy3-preview (High) | 55.9 |
| 3 | DeepSeek V4 Pro (Max) | 54.5 |
| 4 | Seed 2.0 Pro (High) | 54.1 |
| 5 | Gemini 3.5 Flash (High) | 53.9 |
| 6 | Claude Opus 4.7 (Max) | 51.1 |
| 7 | Qwen 3.6 Max (Preview) (Thinking) | 46.5 |
| 8 | Grok 4.3 (High) | 39.4 |
| 9 | gpt-oss-safeguard-20B | 8.8 |
Interactive version: theaggregate.ai/benchmark?slug=safepyramid-l1-rule-dependencies · How It Works · Data refreshed daily, snapshot 2026-09-29.