Google has introduced Gemini 3.5 Flash Cyber, a new artificial intelligence entirely focused on identifying and fixing security vulnerabilities.
Të lidhura
None found
This model arrives as a counter-response to Anthropic’s Gemini Mythos 5, OpenAI’s Gpt 5.5 Cyber, and Microsoft’s Mdash. According to the company’s announcement, Gemini 3.5 Flash Cyber was designed to offer a more economical solution against rival offerings, by minimizing computational expenses. Initially, this tool will be accessible to government institutions and a limited group of partners.
The artificial intelligence has already proven its efficiency by discovering weaknesses within Google’s own codebases, including platforms such as Chrome, Android, Cloud, Ads, and YouTube.
Embedded in the CodeMender security agent, the system operates on a fast, iterative verification cycle. During testing phases, it managed to identify 55 verified vulnerabilities, outperforming general-purpose models and bringing to light critical paths that had previously remained hidden.
On the CyberGym ranking, one of the leading benchmarks for cybersecurity, Gemini 3.5 Flash Cyber surpasses Mythos Preview, but still trails other major competitors. However, its advantage lies in significantly lower cost, which will pave the way for a considerably larger number of companies to utilize it once it is launched. “Artificial intelligences have managed to find security gaps faster than our systems can close them,” Google emphasized in a statement. “To counter this growing risk, a software security strategy is needed that is both powerful and highly efficient.” The company added that this model will be part of a “limited-access pilot program.”
