LLM Trustworthy - Adversarial Demo: leaderboard
The "Adversarial Demo" subtask of LLM Trustworthy. AI-Secure's LLM trustworthiness leaderboard (DecodingTrust) — safety and reliability measured across adversarial robustness, ethics, fairness, out-of-distribution robustness, privacy, stereotype and toxicity dimensions, each reported as...
Metric: Trust Score (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 26 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o Mini (2024-07-18) | 88.49 | #558 |
| 2 | GPT-4o (2024-05-13) | 88.1 | #325 |
| 3 | GPT-4 (0314) | 77.94 | #574 |
| 4 | Llama 3 8B Instruct | 75.54 | #1129 |
| 5 | Gemini 1.0 Pro | 75.54 | #933 |
| 6 | Claude 2 | 72.97 | #761 |
| 7 | tulu-2-13B | 71.17 | #1404 |
| 8 | zephyr-7B-beta | 68.68 | #1330 |
| 9 | Mistral-7B-OpenOrca | 62.15 | #1389 |
| 10 | tulu-2-7B | 60.49 | #1429 |
| 11 | RedPajama-INCITE-7B-Instruct | 58.51 | #1566 |
| 12 | mpt-7B-chat | 58.25 | #1556 |
| 13 | vicuna-7B-v1.3 | 57.99 | #1507 |
| 14 | Llama 2 7B Chat | 55.54 | #1499 |
| 15 | gemma-2B (IT) | 35.55 | #1559 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=llm-trustworthy-adversarial-demo · How It Works · Data refreshed daily, snapshot 2026-10-20.