Introducing Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber

Google is expanding its Gemini lineup with three new models designed for developers building production-grade AI agents. The latest additions include Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber. These releases focus on balancing efficiency, latency, and reliability for high-volume workflows.

Gemini 3.6 Flash serves as the primary update for coding and complex knowledge tasks. Independent benchmarks show a 17 percent reduction in output token usage compared to its predecessor. Beyond the efficiency gains, it features improved computer use capabilities, allowing it to handle multi-step workflows with fewer errors and tool calls. It is now available to developers through Google AI Studio and Android Studio.

For high-throughput requirements, Gemini 3.5 Flash-Lite offers increased speed at a lower cost. It hits 350 output tokens per second, making it suitable for tasks like agentic search and automated document processing. Despite its classification as a lightweight model, it outperforms several previous iterations of the Flash series across coding and long-context benchmarks.

Finally, the company introduced 3.5 Flash Cyber, a specialized model integrated into the CodeMender platform. This tool identifies and patches security vulnerabilities. Due to the sensitive nature of cybersecurity defense, access to this specific model remains restricted to governments and selected partners via a pilot program.

Looking ahead, development on the next generation of models is already underway. Teams are currently engaged in pre-training for Gemini 4 while continuing to test Gemini 3.5 Pro for future release. Developers can access the new models immediately through the Gemini API and enterprise platforms.