CyberSecEval2 Prompt Injection — leaderboard
Metric: Accuracy (%). Source: github.com. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V3 Chat | 48.01 |
| 2 | Llama 3.2 90B Vision Instruct | 37.35 |
| 3 | Llama 3.3 70B Instruct | 36.45 |
| 4 | Grok 2 (1212) | 29.18 |
| 5 | mistral-large-latest | 28.88 |
| 6 | mistral-small-latest | 28.59 |
| 7 | GPT-4o Mini (2024-07-18) | 25.8 |
| 8 | Gemini 2.0 Flash (001) | 24.7 |
| 9 | Claude 3.7 Sonnet (20250219) | 17.43 |
| 10 | GPT-4o (2024-08-06) | 17.13 |
Interactive version: theaggregate.ai/benchmark?slug=cyberseceval2-prompt-injection · How the rankings work · Data refreshed daily, snapshot 2026-07-22.