CyberSecEval2 Vulnerability Exploit — leaderboard

Metric: Accuracy (%). Source: github.com. 11 models tracked.

Top models

#ModelScore
1Claude 3.7 Sonnet (20250219)56.82
2GPT-4o (2024-08-06)53.28
3Gemini 2.0 Flash (001)50.38
4Grok 2 (1212)50.16
5Llama 3.3 70B Instruct48.93
6DeepSeek V3 Chat46.13
7Llama 3.2 90B Vision Instruct40.45
8mistral-large-latest38.11
9O4 Mini (2025-04-16)37.3
10mistral-small-latest35.91
11GPT-4o Mini (2024-07-18)34.15

Interactive version: theaggregate.ai/benchmark?slug=cyberseceval2-vulnerability-exploit · How the rankings work · Data refreshed daily, snapshot 2026-07-22.