In brief: MAI-Cyber-1-Flash handles up to 90 percent of vulnerability detection tasks independently and enables, in combination with other models, cost savings of 50 percent compared to previous MDASH configurations.
Microsoft has introduced MAI-Cyber-1-Flash, its first cybersecurity-specific AI model. The model is designed to achieve a detection rate of approximately 96 percent in the CyberGym test framework and is integrated into Microsoft’s multi-agent framework MDASH.
Microsoft has introduced MAI-Cyber-1-Flash, the company’s first AI model developed specifically for cybersecurity. The model is integrated into MDASH, Microsoft’s multi-agent framework for automated vulnerability detection and remediation. According to the company, MDASH coordinates over 100 specialized AI agents across multiple frontier and specialized models and has already been deployed to uncover vulnerabilities in Microsoft’s own code.
In the cybersecurity test framework CyberGym, the combination of MAI-Cyber-1-Flash, MDASH, and GPT-5.4 achieved a success rate of approximately 96 percent in vulnerability detection. For comparison, according to Microsoft, Google’s 3.5 Flash Cyber, OpenAI’s GPT-5.6 Sol, and Anthropic’s Mythos 5 perform worse; Mythos 5 falls approximately 12 percentage points below that.
MAI-Cyber-1-Flash’s architecture is designed for cost efficiency: the model is intended to handle up to 90 percent of all tasks independently. For the remaining 10 percent of particularly complex tasks, MDASH falls back on larger, more expensive models such as GPT-5.4. This tiered combination of multiple models enables cost savings of 50 percent compared to the previous MDASH configuration consisting of GPT-5.4, GPT-5.4-mini, and GPT-5.3-Codex.
MAI-Cyber-1-Flash is provided via Project Perception, Microsoft’s agent-based security offering. With Project Perception, organizations can simulate attacks, detect and investigate threats, and remediate vulnerabilities. The offering is scheduled to enter public preview starting August 3. Microsoft promises visibility across identities, endpoints, applications, data, clouds, and AI systems.
Source: www.it-daily.net · Published July 28, 2026
Lumi AI News — AI-assisted curation in accordance with Article 50 EU AI Act. Paraphrasing and classification by Lumi News Pipeline v1.7.3.