Cogent Introduces First Mythos-Class Frontier AI Model Built for Cybersecurity Operations

Cogent, a provider of industrial cybersecurity solutions, has unveiled VR-1, a foundation model developed specifically for cybersecurity applications, as the company looks to address growing demand for AI tools tailored to security operations rather than general-purpose use. Cogent trained VR-1 to work the way a skilled adversary does on IntrusionBench, a new benchmark for real-world enterprise attack chains released alongside the model.

The VR-1 was trained on cybersecurity-focused data and is designed to assist with tasks including threat detection, malware analysis, vulnerability assessment, incident response and security research. Unlike general-purpose large language models, the model is intended to provide more specialized capabilities for security professionals by emphasizing domain-specific knowledge and workflows.

Cogent said VR-1 can be integrated into security operations centers and enterprise security platforms to help analysts investigate threats, automate routine tasks and improve response times while keeping human analysts responsible for operational decisions. Cogent is making the model available only to vetted organizations through the Cogent Frontier Access Program, with the company's safety guardrails, policy controls, and audit logging in place.

Qualified participants can also receive a Frontier Model Risk Assessment, a grounded readout of the attack paths a frontier-level model could reach inside their environment, which weaknesses become more dangerous as AI capabilities advance, and where AI can be deployed safely to reduce that risk.

Posted by Pure AI Editors on 07/27/2026