Gemini 3.8 Flash Brings Cyber Chops to Cheap Inference
Google DeepMind's third Flash release in six weeks pairs a sharper reasoning engine with a defender-only cybersecurity variant.
By The ai-compute Desk
Google DeepMind shipped Gemini 3.8 Flash Cyber. Gemini 3.8 Flash is priced identically to 3.7 Flash at $0.75 per million input tokens and $3.75 per million output tokens. [1]
The standard 3.8 Flash targets long-horizon software engineering and autonomous agents. On the DeepSWE v1.1 benchmark it outperforms most larger frontier models while costing a fraction as much, and scores 54.9% on HLE-Verified — a multi-step reasoning suite spanning STEM, humanities, and professional fields. [1]
3.On an internal benchmark spanning 20 programming languages, the model finds vulnerabilities across complex codebases at a success rate exceeding 70%. On CWE-Bench it patches at a pass@1 of 47.2% — within a whisker of the leading frontier model at 47.8% — at significantly lower cost. [1]
Real deployments sharpen the picture. Google's Chrome Security team recorded 2.6 times more correct patches than the best commercial models that are much larger, while the Cloud Vulnerability Research team leveraged the 3.8 Flash Cyber model to find a critical foundational vulnerability in less than 2 hours — work that normally takes months. [1]
The split release illustrates a familiar tension in capable AI: the same core intelligence that accelerates defenders can be pointed the other way. Constraining access at the model tier rather than the application layer is one answer — though whether program-gated access holds under pressure remains the harder question.
3.8 Flash is available in the Gemini API via Google AI Studio and Android Studio, in Stitch, in Gemini Enterprise, and to Google AI Pro and Ultra subscribers across the Gemini app, AI Mode in Google Search, and Gemini in Google Sheets. The Cyber variant requires a Fairwind Program application. [1]
We focused specifically on equipping defenders with expert capabilities that give them an advantage over attackers. This is why we have invested in vulnerability fixing from the start, and prioritized it over offensive capabilities like exploitation.Google DeepMind