id: 3288
uid: b8poX
insdate: 2026-05-29 20:05:06
title: CVE-Bench: testing LLM agents on real-world vulnerability patches
additional: 🤖 CVE-Bench: Testing LLM Agents on Real-World Vulnerability Patches

CVE-Bench is a benchmark designed to evaluate the performance of Large Language Models (LLMs) in identifying and patching real-world vulnerabilities. This practical tool helps assess the ability of LLMs to understand and address cybersecurity threats. By testing LLMs on actual vulnerability patches, CVE-Bench provides valuable insights into their effectiveness.
category: Hacker News
md5:
guid: https://news.ycombinator.com/item?id=48328088
source_url: https://giovannigatti.github.io/cve-bench/
updated:
image:
author_name: logickkk1
author_link: