Google has released Gemini 3.8 Flash and Gemini 3.8 Flash Cyber, splitting its new model family between agentic work and restricted cyber defense. The third Flash launch in six weeks arrives three weeks after 3.7 Flash. Both variants share foundational intelligence, and 3.8 Flash keeps the same introductory pricing: $0.75 per million input tokens and $3.75 per million output tokens.
Gemini 3.8 Flash targets long-horizon coding, autonomous agents, and multi-step reasoning. Google says it beats most larger frontier models on DeepSWE v1.1 and scores 54.9% on HLE-Verified, with further gains on finance and legal agent benchmarks. Google attributes those results to additional reasoning steps and iterative tool calls. That can use more tokens at higher effort levels, so developers can select lower effort or stay on supported 3.7 Flash when compute efficiency matters.
The model is available now through the Gemini API in Google AI Studio and Android Studio, plus Google Antigravity and Stitch. Enterprises can access it in Gemini Enterprise. Google AI Pro and Ultra subscribers can use it in the Gemini app, AI Mode in Search, and Gemini in Google Sheets.
Gemini 3.8 Flash Cyber focuses on vulnerability discovery and automated patching. Google reports a success rate above 70% on an internal benchmark covering complex codebases in 20 languages. Its 47.2% pass@1 result on CWE-Bench trails a leading frontier model’s 47.8% at lower cost. Chrome’s security team says it produced 2.6 times as many correct patches as the best commercial models tested. Wiz measured higher recall at lower cost, while Google’s Cloud Vulnerability Research team used it to find a critical foundational flaw in under two hours.
Google says both releases were accelerated by long-running agentic loops that recursively evaluate and refine the models, with training advances that included cybersecurity work. Google limits cyber access to trusted government authorities, critical-infrastructure operators, and software maintainers through its new Fairwind Program. The Cyber variant uses more permissive cybersecurity mitigations, while the main Flash model carries safeguards for chemical, biological, radiological, nuclear, and cyber-offense risks under Google’s Frontier Safety Framework. The company also reports stronger prompt-injection robustness in Gray Swan testing.