Google announces new AI security model for vulnerability detection
Gemini 3.5 Flash Cyber offers a cost-effective alternative to larger, more expensive models.
1Google's AI-powered security move
Google unveiled a new AI security model called Gemini 3.5 Flash Cyber. This model aims to provide a more cost-effective alternative compared to large, expensive systems. Specializing in quickly detecting code vulnerabilities, the model will initially be available to governments and trusted partners.
2Advantages over competitor models
Gemini 3.5 Flash Cyber offers significant advantages over competitors like Anthropic's Mythos 5. While Mythos reportedly costs twice as much as Claude Opus 4.8, Google's model promises similar performance at lower costs. As companies like Microsoft adopt Mythos, Google is intensifying competition in this space.
3CodeMender integration
Google's security-focused coding agent CodeMender will work integrated with the new model. The system can invoke 3.5 Flash Cyber multiple times at high speed and low cost, enabling scanning of more code paths. This approach promises more comprehensive results in vulnerability detection.
4Performance metrics
Google stated its model demonstrated 'competitive performance' against 'significantly larger models' on the CyberGym AI cybersecurity benchmark. Detecting 55 unique issues in the V8 JavaScript Engine, the model outperformed competitors. It also discovered 10 new vulnerabilities that other models missed.
5Other Gemini updates
The company also announced that Gemini 3.6 Flash offers improvements in coding and multimodal performance, while 3.5 Flash-Lite was introduced as the series' 'most cost-effective' model. These updates are part of Google's diversification strategy in the AI field.
6AI race in cybersecurity
Google's move underscores the growing importance of AI-based solutions in cybersecurity. Microsoft's adoption of Mythos and claims from competitors like China's Z.ai demonstrate intense industry competition. Google's model has potential to stand out in this race with its cost advantages.
Original article: The Verge — Google, güvenlik açıklarını tespit eden yeni bir AI modeli duyurdu