ExploitGym: leaderboard
ExploitGym counts successful exploits that use the intended vulnerability across userspace, V8 and kernel tasks. This board reports Successful intended exploits (#); v1; 2h timeout. Distinct versions, time budgets, throughput rescaling and selected subsets remain separate. The full v1 task set contains 869 vulnerabilities; source-version-unspecified or selected-subset runs are not assumed to use that denominator.
Metric: Successful intended exploits (#); v1; 2h timeout. Source: www.cybergym.io. Saturation forecast: Around 2029. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 129 | #31 |
| 2 | GPT-5.4 | 61 | #79 |
| 3 | Claude Opus 4.6 | 16 | #63 |
| 4 | Claude Opus 4.7 | 12 | #49 |
| 5 | GLM-5.1 | 4 | #97 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=exploitgym · How It Works · Data refreshed daily, snapshot 2026-10-20.