Microsoft Debuts MAI-Cyber-1-Flash Cybersecurity AI

Microsoft launched MAI-Cyber-1-Flash, a cybersecurity-focused AI model, and integrated it into MDASH, the company’s multi-agent orchestration system that manages more than 100 specialized AI agents.

According to Microsoft, the model was tested within the CyberGym vulnerability discovery framework. When MAI-Cyber-1-Flash ran in MDASH alongside GPT-5.4, it outperformed Google’s 3.5 Flash Cyber, OpenAI’s GPT-5.6 Sol and Anthropic’s Mythos 5 for finding vulnerabilities.

Microsoft said: “MAI-Cyber-1-Flash was designed to efficiently handle up to 90% of all tasks, enabling MDASH to use the larger and most costly models in our fleet (in this case GPT-5.4) for the 10% of exceptionally hard tasks that truly need them.”

The company added: “This combination delivers a 50% cost saving when compared against our best offering in MDASH today (GPT 5.4 + 5.4 mini + 5.3 codex). That’s the power of a well-tuned, multi-model system with access to uniquely rich historical training data.”

MAI-Cyber-1-Flash will be offered through Project Perception, an agentic security product that simulates attacks, detects and investigates threats and supports vulnerability remediation. Project Perception is scheduled to enter public preview on August 3.

Microsoft reported that MDASH has been used to find many vulnerabilities in its own codebases and that the company’s historical training data was used to tune the new model. The company described MAI-Cyber-1-Flash as focused on identifying difficult vulnerabilities in complex code while leaving routine tasks to smaller, faster models.

Microsoft said human teams will remain involved to validate findings and carry out remediation based on the models’ outputs.