Google has launched Gemini 3.5 Flash Cyber, a lower-cost AI model designed to find and fix security vulnerabilities, positioning it as an affordable alternative to larger systems such as Anthropic's Mythos. Announced in a blog post on Tuesday, the model is built on Gemini 3.5 Flash and integrates with CodeMender, Google's security-focused coding agent, which can call on it repeatedly at high speed and low cost to scan more code for flaws. The move matters because it signals a widening race among AI firms to automate vulnerability detection, an area currently dominated by expensive, compute-heavy models.

Anthropic's Mythos 5, part of its Project Glasswing initiative, costs twice as much to run as Claude Opus 4.8, and Microsoft's adoption of it reportedly contributed to its biggest Patch Tuesday this month. China's Z.ai has also claimed a model competitive with Mythos. Google says 3.5 Flash Cyber achieved "competitive performance" against much larger models on the CyberGym benchmark when invoked up to five times, and found 55 confirmed issues in the V8 JavaScript engine compared with 47 for Gemini 3.5 Flash and 36 for Opus 4.6, including 10 issues no other model detected. The model will initially be available to governments and trusted partners.