Google Launches Gemini 3.8 Flash AI Model with Dedicated Cybersecurity Variant

On September 2, 2026, Google officially launched Gemini 3.8 Flash alongside a dedicated security variant, Gemini 3.8 Flash Cyber. Marking the company’s third Flash-tier update within six weeks, the new artificial intelligence model aims to deliver higher reasoning and execution capabilities for autonomous AI agents and software engineering tasks at a significantly lower operational cost. The standard Gemini 3.8 Flash model features a 1-million-token context window and speeds reaching up to 305 tokens per second. Google engineered the model to execute extra internal reasoning steps when solving complex, multi-step technical problems. Early benchmarks demonstrate notable performance gains in long-horizon coding, terminal commands, and tool usage compared to its predecessor, Gemini 3.7 Flash, while matching the capabilities of higher-cost frontier models in several key benchmarks. Alongside the general model, Google introduced Gemini 3.8 Flash Cyber, a specialized variant tuned specifically for cybersecurity applications. Available exclusively to verified security defenders through Google’s new Fairwind Program, the cyber model achieves over 70 percent accuracy in real-world vulnerability discovery across 20 programming languages and outperforms existing commercial models in vulnerability patching. The double release underscores Google's strategy to rapidly deploy high-speed, cost-effective AI models to maintain a competitive edge against enterprise rivals like OpenAI and Anthropic. Introductory pricing for developers remains fixed at $0.75 per million input tokens and $3.75 per million output tokens through the end of 2026, lowering barriers for enterprise deployment of autonomous AI workflows.