Cogent's VR-1.1 repairs 57.9% of LiveCVEBench tasks in company run

Cogent Security released VR-1.1 on November 3rd, 2026, a security reasoning model tuned for vulnerability repair, and reported that it fixed 57.9% of 190 LiveCVEBench tasks in the company's own run, 12.6 percentage points ahead of Claude Code and Codex at 45.3% each. Founders Vineet Edupuganti and Geng Sng built Cogent around the gap between fast AI-driven vulnerability discovery and slower patching, betting that a trustworthy fix is the scarce step.

Cogent's VR-1.1 repairs 57.9% of LiveCVEBench tasks in company run Vineet Edupuganti and Geng Sng's Cogent Security reports that VR-1.1 repaired 57.9% of 190 LiveCVEBench tasks, 12.6 percentage points ahead of Claude Code and Codex in its own run. By Ryan Merket https://runtimewire.com/author/ryan-merket · Published Primary source: PR Newswire Business Technology https://www.prnewswire.com/news-releases/cogent-unveils-vr-1-1--its-latest-frontier-cyber-model-which-tops-livecvebench-for-vulnerability-repair-302927966.html Why it matters AI-driven disclosure is producing vulnerabilities faster than engineers can safely patch them, and Cogent's founders are betting the scarce step is a trustworthy fix. Vineet Edupuganti and Geng Sng built Cogent Security around a bottleneck they watched at Abnormal Security: finding vulnerabilities was getting faster, and fixing them was not. On November 3rd, 2026, Cogent released VR-1.1, a version of its security reasoning model tuned for repair, and reported in Cogent's LiveCVEBench research post https://www.cogent.com/blog/vr-1.1-on-livecvebench-remediating-real-vulnerabilities?ref=runtimewire that it fixed 57.9% of 190 tasks, compared with 45.3% for both Claude Code and Codex.… When you purchase through links in our articles, we may earn a small commission https://runtimewire.com/affiliate-monetization-standards . This doesn’t affect our editorial independence.