Google DeepMind launches Gemini 3.8 Flash and Flash Cyber

Google DeepMind launched Gemini 3.8 Flash — its most capable Flash model, it says, for coding and agents — and Gemini 3.8 Flash Cyber for finding and fixing software holes, gated through a Fairwind Program.

Google shipped a cheaper, faster Gemini for coding and a locked cyber variant for trusted defenders. The “most capable Flash” line, the benches, and the price card stay DeepMind’s.

DeepMind published “Introducing Gemini 3.8 Flash and 3.8 Flash Cyber.” It calls 3.8 Flash the most capable Flash model for software engineering and agent work, at the same introductory price as 3.7 Flash.

Gemini 3.8 Flash Cyber is positioned for vulnerability discovery and automated patching — finding software holes and writing fixes. Access is through a new Fairwind Program for trusted defenders — government, critical infrastructure, and software maintainers, per the post. It is not a general API model.

Introductory API pricing is $0.75 per million input tokens and $3.75 per million output tokens through 31 December 2026. From 1 January 2027 the page lists $1.50 / $7.50 — double the introductory rates. Those numbers are DeepMind’s rate card.

3.8 Flash is listed in the Gemini API, Google AI Studio, Android Studio, Gemini Enterprise, and consumer Gemini surfaces (AI Pro and Ultra). Flash Cyber stays Fairwind-gated.

Benchmarks on the launch post and the model card — DeepSWE, CyberGym, CWE-Bench, HLE-Verified — are DeepMind’s scores. Cost comparisons on the page are the lab’s.