CyberSecEval: leaderboard
CyberSecEval: Measures model robustness, truthfulness, calibration, bias, harmfulness, jailbreak resistance, or alignment-relevant behavior.
Metric: Average Injection Success Rate (imported). Source: github.com. Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4 | 19.87 | #438 |
| 2 | Llama 3 70B Instruct | 29.27 | #634 |
| 3 | GPT-3.5 Turbo | 39.13 | #865 |
| 4 | Llama 3 8B Instruct | 45.27 | #1129 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=cyberseceval · How It Works · Data refreshed daily, snapshot 2026-10-20.