Microsoft Says MDASH Beats Claude Mythos and GPT-5.6 Sol in Cybersecurity Test

Microsoft introduced MAI-Cyber-1-Flash, its first dedicated cybersecurity model, and said it is integrated into MDASH, a vulnerability-hunting system that achieved 95.95% on the CyberGym benchmark. Microsoft claims the setup outperformed GPT-5.5 Cyber, Mythos 5, GPT-5.6 Sol and Gemini 3.5 Flash Cyber while cutting cost by 50% versus its current best MDASH configuration. The system uses a hybrid workflow: MAI-Cyber-1-Flash handles up to 90% of tasks, while MDASH escalates the hardest cases to GPT-5.4. The company is positioning the tool as a more efficient way to find, validate and fix software vulnerabilities, with private preview access now available in Microsoft Defender. The announcement is strategically important for Microsoft’s AI and security businesses, reinforcing its push to monetize AI in enterprise cybersecurity and potentially strengthening Defender’s competitive appeal.