Microsoft AI released MAI-Cyber-1-Flash, a 5-billion-parameter cybersecurity model that pushes its MDASH agentic security system to a 95.95% score on the CyberGym benchmark, the company announced July 28.
The model uses a mixture-of-experts architecture with 5 billion active parameters, designed for speed and efficiency in security operations. It is specialized for vulnerability discovery, exploit analysis and defensive code generation, complementing Microsoft’s broader push into AI-powered cybersecurity tools.
MAI-Cyber-1-Flash builds on the MDASH framework, which orchestrates multiple specialized AI agents that collaborate to discover, debate and verify security bugs. The new model delivers what Microsoft describes as frontier-grade cyber capabilities at a fraction of the computational cost of larger models.
The release comes amid a broader industry push toward dedicated AI cybersecurity models, with firms racing to develop specialized tools that can keep pace with AI-powered threats. Microsoft has positioned AI-powered security as a strategic priority, integrating these capabilities across its Azure and Defender product lines.
Early benchmarks suggest MAI-Cyber-1-Flash outperforms general-purpose models of comparable size on cybersecurity-specific tasks, though independent verification of the CyberGym results is still pending.
