Gemini Flash Cyber: Google's New AI Security Model

Google has launched Gemini 3.5 Flash Cyber, a new AI security model. This model is specifically designed to detect, validate, and patch software vulnerabilities. It aims to provide high-level cybersecurity performance at a lower cost, especially for large codebases. Google states it is more efficient and performs competitively with much larger models. This new model enters the race to build specialized cybersecurity tools for defenders. Heather Adkins, Google's vice president of security, notes that fixing vulnerabilities takes time and resources. Google also debuted Gemini 3.6 Flash, a smaller model for coding, and 3.5 Flash-Lite, which is now the fastest in the Gemini series. The 3.5 Flash Cyber model was built on Gemini 3.5 Flash and powers Google's CodeMender tool. This tool will now use 3.5 Flash Cyber to scan code, build exploits, and generate patches. Benchmarks show 3.5 Flash Cyber performs well against larger AI systems, sometimes finding flaws its competitors missed. This matters because it offers a more affordable and scalable way to address growing cybersecurity threats.