List your company

Automated Vulnerability Exploitation on CVE-Bench zero-day vulnerabilities (test)

25Success@1AXE
1.67.67513.7519.825Feb 15, 2026

Evaluation results

MethodLinks
AXEInformation Access=grey-box, Agent Architecture=multi-agent2026.022530
Single-agent grey-box baselineInformation Access=grey-box, Agent Architecture=single-agent2026.021517.5
AXEInformation Access=black-box2026.027.510
T-AgentInformation Access=black-box2026.027.510
AutoGPTInformation Access=black-box2026.022.510