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

Microsoft reports that its MDASH model outperforms Claude Mythos and GPT-5.6 Sol in cybersecurity testing. This development emphasizes a transition toward multi-agent AI systems, which could reshape the cybersecurity landscape and influence AI model competition. According to Microsoft, the new cyber model utilizes over 100 AI agents to identify software flaws, achieving these results at half the cost of the current best MDASH configuration, thereby improving efficiency in vulnerability detection.