344
Productivity & Workflow355
Automation & Workflow225
Software Development251
Marketing & Growth192
AI Infrastructure & MLOps175
Writing & Content Creation203
Data & Analytics142
Photography & Imaging156
Design & Creative170
Customer Support132
Sales & Outreach125
Voice & Speech135
Education & Learning131
Operations & Admin87
Google launched Gemini 3.8 Flash, plus Gemini 3.8 Flash Cyber for finding software security flaws. It is the third Flash release in six weeks.
In short: Google has launched Gemini 3.8 Flash, plus a separate Gemini 3.8 Flash Cyber model aimed at finding and fixing security bugs.
Google announced Gemini 3.8 Flash, its third “Flash” model release in about six weeks. Flash models are meant to be fast and lower cost than Google’s larger “Pro” models. Google has not released a new top tier Gemini Pro model since early 2026.
Gemini 3.8 Flash comes in two versions. The standard Gemini 3.8 Flash is meant for general tasks, including writing code and handling “agent” tasks, which means the AI tries to complete a multi step job on your behalf (like a helper that follows a checklist). The second version, Gemini 3.8 Flash Cyber, is tuned for cybersecurity work, like spotting weak points in software and suggesting fixes.
For developers who use Google’s API (a way for software to talk to other software, like ordering from a menu), Google is offering discounted pricing through the end of the year. The company lists $0.75 per million input tokens and $3.75 per million output tokens. Tokens are small chunks of text that work like a word count meter.
Google says its own test results show small improvements over Gemini 3.7 Flash in many areas, and bigger gains in coding. It also says the model tops the DeepSWE leaderboard for complex software engineering tasks, but still lags behind rivals in tests where the AI has to operate a computer.
For everyday users, this mainly affects what you can do inside Google’s Gemini app and AI Studio, especially for coding and other structured tasks. For security teams, Google is signaling that AI tools are becoming part of how companies find and patch software flaws faster, even if the Cyber version is currently limited to trusted testers and governments.
Source: Arstechnica