- Google has officially announced its next-generation AI models, “Gemini 3.8 Flash” and the cybersecurity-focused AI model “Gemini 3.8 Flash Cyber.”
- “Gemini 3.8 Flash” is a flagship AI model that maintains the same speed and low cost as “Gemini 3.7 Flash” while significantly improving its capability to handle software development, agent tasks, and complex operations in specialized fields.
- “Gemini 3.8 Flash Cyber” is an AI model specialized in discovering software vulnerabilities and automatically creating patch fixes for them.
Just six weeks after announcing the new “Gemini 3.7 Flash” AI model on Thursday, August 13, 2026, Google officially announced its next-generation AI models, “Gemini 3.8 Flash” and the cybersecurity-focused “Gemini 3.8 Flash Cyber,” on Wednesday, September 2, 2026 local time.
Marking an unusual pace of three “Gemini Flash” model rollouts in just six weeks, both “Gemini 3.8 Flash” and “Gemini 3.8 Flash Cyber” are developed based on the same foundational model, featuring rigorous training in the cybersecurity field along with substantial enhancements in coding and reasoning capabilities.
“Gemini 3.8 Flash”
“Gemini 3.8 Flash” is a flagship AI model that maintains the same speed and low cost as “Gemini 3.7 Flash” while vastly improving its ability to handle software development, agent tasks, and complex operations in specialized domains.
Its capability to autonomously handle time-consuming and complex software development tasks, as well as specialized workflows in fields like finance and law, has been elevated to match or even surpass larger, more costly models. These performance improvements stem from a design where “Gemini 3.8 Flash” itself persistently applies additional reasoning and tool usage for complex tasks.
However, because this can sometimes increase processing time and costs, users who prioritize cost-efficiency can continue using “Gemini 3.7 Flash,” which remains supported. Additionally, “Gemini 3.8 Flash” retains the protective features carried over from “Gemini 3.7 Flash” to prevent misuse in chemical, biological, radiological, nuclear (CBRN), and cyber attack domains.
jetstream.blog
For developers, “Gemini 3.8 Flash” is available through the agent development environment “Google Antigravity,” as well as “Google AI Studio” and “Android Studio.” For enterprise customers, it is offered via “Gemini Enterprise,” and for general consumers subscribed to “Google AI Pro” and “Google AI Ultra,” it is available in the Gemini app, the “AI Mode” search service, and the “Gemini in Google Sheets” feature.
Developer pricing for “Gemini 3.8 Flash” remains at the same level as “Gemini 3.7 Flash,” priced at $0.75 per million input tokens and $3.75 per million output tokens. However, prices are scheduled to double to $1.50 and $7.50, respectively, starting Friday, January 1, 2027.
“Gemini 3.8 Flash Cyber”
“Gemini 3.8 Flash Cyber” is an AI model specialized in detecting software vulnerabilities and automatically generating patch fixes for them.
It is now capable of identifying vulnerabilities with an accuracy that surpasses the previous model, “Gemini 3.5 Flash Cyber,” and even larger-scale models. It is already being utilized within Google, with reported successes including vulnerability fixes by the Chrome security team, high detection accuracy during penetration testing by security firm Wiz, and instances where critical vulnerabilities that typically take months to discover were found in just a few hours.
Due to its characteristics and from a misuse-prevention standpoint, “Gemini 3.8 Flash Cyber” will not be released to the general public, much like “Gemini 3.5 Flash Cyber.” Instead, it will be provided preferentially only to trusted defenders—such as government agencies, critical infrastructure operators, and software maintainers—through a newly established limited-access program for advanced cybersecurity defense called the “Fairwind Program.”
Source：Google (https://blog.google/innovation-and-ai/models-and-research/gemini-models/3-8-flash-and-3-8-flash-cyber/)