Google unveils an affordable option to compete with major AI security models like Mythos.
Google is rolling out Gemini 3.6 Flash along with a new security framework designed to swiftly identify and address security vulnerabilities. In a recent blog update, Google highlighted Gemini 3.5 Flash Cyber as a “cost-efficient and highly capable alternative” to larger, pricier AI systems, like Anthropic’s Mythos.
This cybersecurity model builds on Gemini 3.5 Flash and will initially be accessible to governments and trusted partners through CodeMender, Google’s security-centric coding agent. According to Google, CodeMender can leverage 3.5 Flash Cyber “multiple times at high speed and low cost,” enabling AI agents to explore more code paths and uncover vulnerabilities.
Anthropic’s Mythos 5 is a robust AI security model introduced as part of the company’s Project Glasswing initiative. This compute-intensive model comes with a hefty price tag, costing twice as much as Claude Opus 4.8. Microsoft, which has integrated Mythos for its security assessments, experienced its largest Patch Tuesday this month after utilizing AI to detect vulnerabilities. Other companies, including Google, are striving to keep pace, while China’s Z.ai asserts that its model can rival Mythos.
In addition to 3.5 Flash Cyber, Google has announced that the upgraded Gemini 3.6 Flash brings enhancements in coding and multimodal performance, while the new 3.5 Flash-Lite is touted as the “most cost-effective” option in the 3.5 lineup.
Furthermore, Google reports that 3.5 Flash Cyber demonstrated “competitive performance” against “significantly larger models” on the CyberGym AI cybersecurity benchmark when invoked up to five times. It identified 55 “unique confirmed issues” in the V8 JavaScript Engine, surpassing the 47 found by Gemini 3.5 Flash and the 36 detected by Opus 4.6. Notably, 3.5 Flash Cyber uncovered 10 issues that no other model had detected, indicating its ability to continue discovering new code paths and vulnerabilities after multiple invocations.
