Google introduced the Gemini 3.8 Flash and Gemini 3.8 Flash Cyber models, designed for programming tasks and vulnerability detection, respectively.
Both artificial intelligence models share the same technological foundation and are geared towards performing autonomous work cycles that evaluate and refine their responses. Google has indicated that the improvements in programming are also based on training in cybersecurity.
Gemini 3.8 Flash is available to developers through Google AI Studio, Android Studio, Gemini API, Google Antigravity, and Stitch. Companies can access it via Gemini Enterprise, while subscribers of Google AI Pro and Ultra can use it in the Gemini app, Google Search’s AI mode, and Gemini in Google Sheets. Meanwhile, Gemini 3.8 Flash Cyber will be available to defenders through Fairwind, a limited access program for governments and trusted partners.
Gemini 3.8 Flash is Google’s most advanced model for software engineering and complex tasks. This model, which maintains the speed and pricing of Gemini 3.7 Flash, costs USD 0.75 per million input tokens and USD 3.75 per million output tokens. Additionally, it has demonstrated superior performance in several assessment tests, such as DeepSWE v1.1 and Vals Finance Agent V2.
On the other hand, Gemini 3.8 Flash Cyber focuses on vulnerability detection and offers code correction capabilities. Google has prioritized defensive features, reporting that the model achieved a success rate of over 70% in an internal assessment on codebases. Moreover, the Chrome security team reported that this model generated 2.6 times more correct vulnerability fixes compared to other commercial models.
These launches underscore Google’s commitment to advancing artificial intelligence to enhance both programming efficiency and cybersecurity in today’s digital environment.
Source: www.infobae.com
