Harbor Adapters - StrongREJECT (Terminus-2): leaderboard
Metric: Defense score (%; 150 prompts, 100 minus the StrongREJECT score, mean of 3 trials). Source: arxiv.org. Saturation forecast: Estimated already saturated. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Haiku 4.5 | 97 |
| 2 | GPT-5 Nano | 96.1 |
| 3 | GPT-5.4 (Medium) | 95.9 |
| 4 | GPT-5 Mini | 93.5 |
| 5 | Claude Opus 4.6 | 91 |
| 6 | Claude Sonnet 4.6 | 90.9 |
| 7 | Gemini 3.1 Pro (Preview) | 86.8 |
| 8 | Gemini 3 Flash (Preview) | 73.3 |
Interactive version: theaggregate.ai/benchmark?slug=harbor-adapters-strongreject-terminus-2 · How It Works · Data refreshed daily, snapshot 2026-09-25.