Google Launches Cheaper AI Security Rival to Mythos
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Google launched Gemini 3.5 Flash Cyber as a "cost-efficient and highly capable alternative" to larger systems like Anthropic's Mythos 5, making it available first to governments and trusted partners via the CodeMender coding agent.
- Gemini 3.5 Flash Cyber identified 55 unique confirmed issues in the V8 JavaScript Engine, compared to 47 found by Gemini 3.5 Flash and 36 by Opus 4.6, including 10 vulnerabilities that no other model discovered.
- Anthropic's Mythos 5 costs twice as much as Claude Opus 4.8 to run, and Microsoft adopted Mythos for security checks, producing its biggest Patch Tuesday this month after AI-discovered vulnerabilities.
- CodeMender can invoke 3.5 Flash Cyber "multiple times at high speed and low cost," letting AI agents scan more code paths than compute-heavy alternatives, and the model achieved "competitive performance" against significantly larger models on the CyberGym AI cybersecurity benchmark.
- Google also rolled out an upgraded Gemini 3.6 Flash with coding and multimodal improvements, plus Gemini 3.5 Flash-Lite as the "most cost-effective" model in the 3.5 series.
- China's Z.ai claims its model can compete with Mythos, adding a third major entrant to the AI security race alongside Google and Anthropic.
Why it matters: Google is undercutting Anthropic's Mythos 5 — which costs twice as much as Claude Opus 4.8 — with a security model that can be invoked repeatedly at low cost. The 3.5 Flash Cyber model also found 10 unique V8 vulnerabilities no other model caught, suggesting cheaper doesn't mean weaker in the AI security arms race that now includes China-backed Z.ai.
