Google launches Gemini 3.8 Flash: third model in six weeks

Google launched Gemini 3.8 Flash and Gemini 3.8 Flash Cyber, its third Flash model in six weeks, targeting complex reasoning, software development, and automated cybersecurity. The models are priced at $0.75 per million input tokens and $3.75 per million output tokens, with Flash Cyber achieving over 70% success in finding security flaws across 20 programming languages. Google is rolling out Gemini 3.8 Flash immediately to developers and business users, while Flash Cyber is limited to trusted partners in the public sector, software industry, and critical infrastructure.

Google is expanding its Gemini range with Gemini 3.8 Flash and Gemini 3.8 Flash Cyber. The new models focus on complex reasoning tasks, software development, and automated cybersecurity. Google is bringing its latest AI models to market with Gemini 3.8 Flash https://blog.google/innovation-and-ai/models-and-research/gemini-models/3-8-flash-and-3-8-flash-cyber/ and 3.8 Flash Cyber. This marks the third Flash rollout in six weeks, just three weeks after the arrival of Gemini 3.7 Flash https://itdaily.com/news/software/google-launches-gemini-3-7-flash-for-improved-encoding-tasks/ . Both models share the same technological foundation but are tailored to different business use cases and environments. Gemini 3.8 Flash specifically targets software development, long-term coding tasks, and workflows with autonomous AI agents. The price remains unchanged at $0.75 per million input tokens and $3.75 per million output tokens, which is the same cost as the previous generation. Better performance for complex tasks According to Google, Gemini 3.8 Flash delivers clear performance gains in autonomously solving technical issues. In software development benchmarks, the model outperforms several larger alternatives. The model also records higher scores within specialized fields such as financial analysis and legal support. The progress stems from an approach where the model goes through more reasoning steps and repeatedly calls tools. Developers who primarily want to save computing power can lower the effort level or continue to rely on Gemini 3.7 Flash for more efficient workloads. Flash Cyber for defenders With Gemini 3.8 Flash Cyber, the tech giant introduces a variant specifically trained for cybersecurity. This model focuses primarily on detecting vulnerabilities and automated writing of security patches. On internal benchmarks with codebases in twenty programming languages, the model achieves a success rate of over seventy percent in finding security flaws. Google is now using Flash Cyber itself to secure internal code. For instance, the Google Cloud security team discovered a critical error in less than two hours that might otherwise have gone unnoticed. Subsidiary Wiz https://itdaily.com/news/business/google-acquires-wiz/ also reports better detection results at lower operational costs. Availability and rollout Google is making Gemini 3.8 Flash available immediately for developers via Google AI Studio and Android Studio, and for business users within Gemini Enterprise. Paying consumers with an AI Pro or Ultra subscription also gain access to the model via the Gemini app and Google Search. Gemini 3.8 Flash Cyber, on the other hand, has a more limited distribution. To prevent misuse, this model is only offered to trusted partners from the public sector, software industry, and critical infrastructure. Launch after launch Google is significantly picking up the pace of its model launches. In just six weeks, we went from Gemini Flash 3.5 to Flash 3.8. Notably, Google is focusing on the smaller, cheaper Flash models. A new Gemini Pro model was also expected to launch in the summer, but the wait for that continues.