Google launches Gemini 3.8 Flash and 3.8 Flash Cyber

Building on the rapid momentum of the 3.7 Flash version, Google has officially unveiled its latest generation of large language models: Gemini 3.8. This marks the launch of the third Flash version in just six weeks, reinforcing Google's leadership in the race for high-performance, cost-optimized AI.

Gemini 3.8 is Google's most capable model to date in terms of reasoning and coding, boasting extremely fast processing speeds and very low costs, similar to version 3.7. This launch includes two specialized variants: Gemini 3.8 Flash and Gemini 3.8 Flash Cyber.

Breakthrough in shared core architecture

Although tailored for different deployment environments, both versions of Gemini 3.8 operate on the same core artificial intelligence platform. This breakthrough is driven by continuous agentic loops—a mechanism that automatically evaluates and recursively refines the foundation models.

The remarkable strides in the programming and multi-step reasoning capabilities of the core system are the result of rigorous training, particularly within the challenging environment of the cybersecurity domain.

Gemini 3.8 Flash: Optimized for long-context programming tasks and autonomous agents (Agentic AI)

Gemini 3.8 Flash delivers significant improvements over version 3.7 Flash, often achieving performance approaching that of leading, higher-cost models.

For DeepSWE v1.1 (Long-horizon Software Engineering), the Flash 3.8 version outperforms most larger-scale frontier models in autonomously solving complex engineering problems from end to end, at a fraction of the cost.

Furthermore, 3.8 Flash demonstrates the reliability essential for critical enterprise autonomous operations across specialized domains. In quantitative and professional fields requiring in-depth analysis and reporting, 3.8 Flash outperforms 3.7 Flash and other advanced models in benchmarks such as Vals Finance Agent V2 and vey’s Legal Agent Benchmark. It also achieves a score of 54.9% on HLE-Verified, highlighting its capability for multi-step reasoning across STEM, the social sciences and humanities, and specialized professional areas.

These performance improvements stem from a core design choice: 3.8 Flash operates with greater intensity. For complex tasks, the model demonstrates greater thoroughness—performing additional reasoning steps and iterative tool calls. At times, the model may utilize more tokens to maximize performance, particularly at higher levels of effort.

Gemini 3.8 Flash Cyber: A state-of-the-art cybersecurity shield.

Exclusively provided to trusted defense entities through the Fairwind program, Gemini 3.8 Flash Cyber ​​delivers a decisive defensive advantage through ultra-fast processing speeds and cost-efficiency, enabling the implementation of continuous cybersecurity enhancement processes.

1. Automatically detect security vulnerabilities on a large scale.

It outperforms both its predecessor, 3.5 Flash Cyber, and other large-scale commercial models in industry-standard tests. To address real-world defense needs extending beyond standard C/C++ codebases, Google evaluated the model using a complex internal testing system spanning 20 programming languages, achieving a vulnerability detection rate of over 70%.

2. Automated patching (Patching vulnerabilities before an attack occurs)

Google places a higher priority on vulnerability remediation (defensive) capabilities than on offensive (exploitation) capabilities. On the CWE-Bench leaderboard (operated by Collinear), Gemini 3.8 Flash Cyber ​​occupies an optimal position on the Pareto frontier: it achieves a pass@1 rate of 47.2%—closely trailing the 47.8% achieved by the leading frontier model—yet operates at a significantly lower cost.

Secure Design and Prompt Injection Resistance

• Frontier Safety: Both models come equipped with stringent safeguards against misuse in high-risk areas such as Chemical, Biological, Radiological, and Nuclear (CBRN) domains and cybercrime.

• Prompt Injection: The Gemini 3.8 models demonstrate a significant leap in resilience against malicious prompt injection attacks, as measured and certified by Gray Swan.

Get started today

1. Developers

• Experience the agent-first workflow in Google Antigravity.
• Integrate and call the API via Google AI Studio or connect directly within Android Studio.
• Rapidly design user interfaces in Stitch.
• Refer to the detailed guide in the Gemini API developer documentation.

2. For Enterprises
Administrators and enterprises can directly access the Gemini 3.8 Flash model within Gemini Enterprise (Google Cloud Console).

2. For End-users
Now available to users subscribed to Google AI Pro and Ultra plans across the entire Workspace ecosystem: the Gemini app, AI Mode in Google Search, and Gemini in Google Sheets.

4. Dành cho Tổ chức An ninh mạng & Chính phủ
Các cơ quan chính phủ đáng tin cậy, đơn vị vận hành hạ tầng trọng yếu và các nhà phát triển duy trì phần mềm nguồn mở có thể Đăng ký quyền truy cập sớm vào Gemini 3.8 Flash Cyber thông qua Chương trình Fairwind.

Master cutting-edge AI technology with Gimasys.

The launch of Gemini 3.8 Flash and 3.8 Flash Cyber ​​marks a major shift for enterprises toward the era of agentic AI and AI-driven proactive defense. However, effectively integrating these ultra-fast AI models into the right business and security workflows requires significant expertise in system design and risk management.

As a premier Google Cloud strategic partner in Vietnam, Gimasys is a trusted partner helping your business fully harness the power of the new Gemini 3.8 generation:

•  Tư vấn thiết kế Agentic AI: Gimasys assists enterprises in building autonomous agents based on Gemini 3.8 Flash, optimizing automation processes for programming, financial analysis, or long-term legal reporting.
•  Optimizing enterprise cybersecurity: Collaborate with the enterprise security team to deploy cybersecurity AI models, optimize source code inspection and automated patching processes, and mitigate the risk of command injection attacks.
•  Cost Management & Copyright: Assist businesses in optimizing API usage costs for Google AI Studio and Gemini Enterprise by configuring appropriate model effort levels to balance performance and budget.

Is your business ready to lead the way in the era of autonomous AI and intelligent security? Contact Gimasys today to receive in-depth consultation on the deployment roadmap for the Gemini 3.8 generation and comprehensive protection for your technology systems!