Google Unveils Gemini 3.5 Flash Cyber AI for Vulnerability Hunting
The specialized AI model, designed for governments and trusted partners, detects and patches software vulnerabilities.
1Exclusive Security AI
Google DeepMind announced Gemini 3.5 Flash Cyber, a specialized AI model designed to detect and patch software vulnerabilities. Due to dual-use risks, the model will be available only to governments and trusted partners. Access will be limited through a pilot program on the CodeMender platform.
2Fast and Cost-Effective Solution
Gemini 3.5 Flash Cyber is optimized for lower costs compared to larger models. It offers extensive code coverage at a reduced per-token cost. The multi-agent system uses parallel agents to produce a single report, enhancing cost efficiency. This approach significantly improves code coverage per session.
3Pioneering Cyber Threat Defense
Google highlights that AI can now find security vulnerabilities faster than current systems can patch them. This creates a threat that outpaces defense response times. Gemini 3.5 Flash Cyber aims to close this gap by prioritizing defense. The model enables early detection and patching of critical vulnerabilities before exploitation.
4Additional Features and Enhancements
Google plans to expand the model with red-teaming features and enterprise defense capabilities. Gemini 3.6 Flash and 3.5 Flash-Lite were also announced. These models offer better coding and knowledge work performance while improving token efficiency. 3.6 Flash consumes 17% fewer tokens than its predecessor.
5Price and Performance Advantage
Gemini 3.6 Flash is priced lower than 3.5 Flash, at $1.50 per million input tokens and $7.50 per million output tokens. 3.5 Flash-Lite stands out as the fastest model, delivering 350 output tokens per second. It offers an ideal price-performance ratio for high-throughput production traffic.
6A Milestone in Cybersecurity
Gemini 3.5 Flash Cyber marks a significant advancement in cybersecurity. The model accelerates defense response times, preventing exploitation of critical vulnerabilities. Google's move underscores AI's potential in security. The restricted access strategy minimizes misuse risks, ensuring responsible deployment.