Cogent AI Team Releases VR-1: A Frontier Cyber Reasoning Model That Composes and Verifies Enterprise Attack Paths

Cogent AI launched VR-1, a cybersecurity reasoning model designed to simulate and verify enterprise attack paths. It includes IntrusionBench, a benchmark for intrusion completion, and is available to vetted large enterprises under strict controls. VR-1 focuses on investigating, composing evidence, recovering from dead ends, and verifying objectives within complex environments.