Crimson AI NewsA CrimsonLingua Network service
EN ع
← Back to news
Google DeepMind

Google DeepMind Unveils Gemini 3.8 Flash and Cyber for Agentic AI and Security

AI By Crimson AI Google DeepMind 2 September 2026 · 16:18 42 views
Share: X Telegram

Google DeepMind introduces Gemini 3.8 Flash, its most intelligent workhorse model, and Gemini 3.8 Flash Cyber, a specialized cybersecurity model, both at Flash speed and cost.

Google DeepMind Unveils Gemini 3.8 Flash and Cyber for Agentic AI and Security

Key points

Google DeepMind has announced the release of Gemini 3.8 Flash and Gemini 3.8 Flash Cyber, marking the third Flash release in just six weeks. Building on the momentum of the previous 3.7 Flash, the new models aim to deliver next-generation intelligence for agentic workflows and cybersecurity while maintaining the same speed and low cost as its predecessor.

Gemini 3.8 Flash is positioned as the company's most intelligent workhorse model, offering significant improvements over 3.7 Flash across software engineering, agentic tasks, and multi-step reasoning in specialized domains. It is available at the same introductory price as 3.7 Flash: $0.75 per million input tokens and $3.75 per million output tokens. The model excels in long-horizon coding and autonomous agent tasks, often approaching the performance of higher-cost frontier models, as evidenced by its performance on benchmarks like DeepSWE v1.1 and HLE-Verified (54.9%).

Gemini 3.8 Flash Cyber is designed for cybersecurity professionals, with frontier-level performance in vulnerability detection and automated patching. It is available through the new Fairwind Program, which provides trusted defenders with prioritized access. On the CyberGym benchmark, it surpasses both previous models and larger frontier models in autonomous vulnerability discovery. On an internal benchmark covering 20 programming languages, it achieves a success rate exceeding 70%. For patching, it scores 47.2% pass@1 on CWE-Bench, nearly matching a leading frontier model at 47.8% but at a significantly lower cost.

The models are already being used to secure Google's code. For example, the Chrome Security team found that 3.8 Flash Cyber produced 2.6 times more correct patches than larger commercial models. Wiz reported higher recall on their penetration testing benchmark at lower cost, and Google's Cloud Vulnerability Research team used the model to find a critical vulnerability in less than 2 hours—a task that typically takes months.

Both models are powered by the same foundational intelligence, with innovations including rigorous training in cybersecurity and long-running agentic loops that recursively evaluate and refine the models. Gemini 3.8 Flash is available to developers via the Gemini API, Google AI Studio, and Android Studio, and to consumers through Google AI Pro and Ultra subscriptions. Gemini 3.8 Flash Cyber is available to trusted defenders through the Fairwind Program.

ModelPrice (per million tokens)Key BenchmarkAvailability
Gemini 3.8 Flash$0.75 input / $3.75 outputHLE-Verified: 54.9%Gemini API, AI Studio, Android Studio, Gemini Enterprise, AI Pro/Ultra
Gemini 3.8 Flash CyberSame as FlashCWE-Bench: 47.2% pass@1Fairwind Program (trusted defenders)
Source
Google DeepMind · Google DeepMind
Related news
Google DeepMind
Google DeepMind 14 Oct 2026

Google DeepMind and Singapore Deploy On-Device AI to Fight Scams

Google DeepMind and Singapore's GovTech have built an on-device scam detection model, fine-tuned on local scam conversations, that...

4
Google DeepMind
Google DeepMind 13 Oct 2026

Google DeepMind Details Safety Framework as Gemini 4 Argon Rolls Out Broadly

Google DeepMind is expanding access to Gemini 4 Argon after a limited pilot showed frontier-level cyberdefense capabilities, inclu...

5
Google DeepMind
Google DeepMind 13 Oct 2026

Google DeepMind Unveils Gemini 4 Argon for Robotics, Scoring 87% on Embodied Reasoning Benchmark

Gemini 4 Argon brings a step change in physical understanding, code generation, and visual-semantic reasoning, scoring 87% on the...

6