ELBench - Safety and Trustworthiness: leaderboard
Metric: Module score (%; mean per-item score on 1,000 items: synthesized refusal, safe-guidance and benign-answering requests and curated EduGuardBench teaching-safety and jailbreak items; zero-shot, temperature 0; open-ended items scored by a Qwen3.6 rubric judge, majority of 9 calls). Source: arxiv.org. Saturation forecast: Around December 2026. 9 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V4 Flash | 89.7 |
| 2 | GLM-5.1 | 89.5 |
| 3 | DeepSeek V4 Pro | 86.1 |
| 4 | Seed 2.0 Pro | 83 |
| 5 | Claude Opus 4.8 | 78.5 |
| 6 | GPT-5.4 | 76.6 |
| 7 | Gemini 3.5 Flash | 70.2 |
Interactive version: theaggregate.ai/benchmark?slug=elbench-safety-and-trustworthiness · How It Works · Data refreshed daily, snapshot 2026-09-29.