“AI Disruption” Publication 10,500 Subscriptions 20% Discount Offer Link.
Today, Google officially launched Gemini 3.8. This is the third Flash-series model Google has released in just six weeks.
According to the announcement, Gemini 3.8 Flash maintains the same speed and cost advantages as Gemini 3.7 Flash while further improving reasoning and coding capabilities, with a particular focus on software engineering, agent tasks, and complex multi-step reasoning scenarios.
Gemini 3.8 includes two versions:
Gemini 3.8 Flash: a high-intelligence general-purpose workhorse model. Compared with Gemini 3.7 Flash, it delivers significant gains in software engineering, agent tasks, and critical multi-step reasoning in specialized domains.
The model uses the same launch pricing as Gemini 3.7 Flash: $0.75 per million input tokens and $3.75 per million output tokens.
At the same time, Google also introduced a cybersecurity-specialized model, Gemini 3.8 Flash Cyber.
The model is positioned as one of Google’s strongest cybersecurity models to date, with a focus on vulnerability discovery and automated patch generation. It will be made available to certified defensive security professionals through Google’s Fairwind Program.
Google said that although the two versions are intended for different deployment environments, they are built on the same underlying model capabilities and are further strengthened through long-running agent loop mechanisms.
