VulnLLM-R (plus black-box web app security benchmark): leaderboard
VulnLLM-R (plus black-box web app security benchmark) reports source-table results for models on cybersecurity tasks spanning white-box vulnerability detection and black-box web application security testing. The evaluated model uses code and web-app testing inputs, and the pinned metric is F1 Score, measuring the balance of precision and recall on the benchmark. Source: BenchmarkList’s attributed result table.
Metric: F1 Score. Source: benchmarklist.com. Saturation forecast: Around August 2027. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 0.81 | #64 |
| 2 | Claude Opus 4.7 | 0.77 | #51 |
| 3 | Claude Sonnet 4.6 | 0.74 | #94 |
| 4 | Gemini 3 Flash | 0.71 | #103 |
| 5 | GPT-5.4 | 0.71 | #83 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=vulnllm-r-plus-black-box-web-app-security-benchmark · How It Works · Data refreshed daily, snapshot 2026-11-05.