Google has launched three new Gemini AI models, including the efficient 3.6 Flash, the high-speed 3.5 Flash-Lite, and a specialized cybersecurity model designed to identify and patch software vulnerabilities.

Key Points

• Gemini 3.6 Flash improves coding and multimodal performance while reducing token usage by up to 17% compared to the previous version.
• Gemini 3.5 Flash-Lite offers high-speed output of 350 tokens per second and includes new built-in tools for agentic computer-use tasks.
• The 3.5 Flash Cyber model integrates with the CodeMender infrastructure agent to detect and fix security bugs in platforms like Android, Chrome, and YouTube.
• Google confirmed that the flagship Gemini 3.5 Pro is currently in partner testing, while pretraining for the next-generation Gemini 4 model has already begun.
• Developers can access the new 3.6 Flash and 3.5 Flash-Lite models immediately through the Gemini API, Google AI Studio, and Android Studio.

Why it Matters

These releases demonstrate Google's strategy to prioritize operational efficiency and specialized utility alongside its ongoing development of flagship AI models. By providing targeted tools for cybersecurity and high-speed agentic workflows, the company aims to maintain its competitive edge in an increasingly crowded generative AI market.