CyberGym: leaderboard
Cybersecurity agent benchmark for discovering, exploiting, and reasoning about vulnerabilities in controlled challenge environments.
Metric: Score (%). Source: llm-stats.com. Status: years away from saturation. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GLM-5.3 | 84.5 |
| 2 | DeepSeek V4 Pro (0813) | 83.3 |
| 3 | Claude Mythos Preview | 83.1 |
| 4 | GPT-5.5 | 81.8 |
| 5 | Claude Opus 4.8 | 78.8 |
| 6 | Hy4 preview | 78.4 |
| 7 | DeepSeek V4 Flash (0731) | 76.7 |
| 8 | Claude Opus 4.6 | 73.8 |
| 9 | Claude Opus 4.7 | 73.1 |
| 10 | Seed 2.1 Pro | 70.2 |
| 11 | GLM-5.1 | 68.7 |
| 12 | Seed 2.1 Turbo | 67 |
| 13 | Kimi K2.5 | 41.3 |
Interactive version: theaggregate.ai/benchmark?slug=cybergym · How It Works · Data refreshed daily, snapshot 2026-09-05.