Details
- Google DeepMind announced three new Gemini models aimed at making AI agents faster, smarter and cheaper at scale.
- Gemini 3.6 Flash is positioned as a higher-quality successor to 3.5 Flash, using fewer output tokens while maintaining the same price point for developers.
- Gemini 3.6 Flash targets general-purpose and agentic workloads, including software development, analytical work and multimodal tasks, with optimizations for fewer reasoning steps and tool calls.
- Gemini 3.5 Flash-Lite is introduced as a fast, cost-effective model for everyday and high-volume tasks, such as translation and bulk data processing, with lower pricing and higher output token throughput.
- Both Gemini 3.6 Flash and 3.5 Flash-Lite are rolling out in the consumer Gemini app and are accessible to developers via the Gemini API in Google AI Studio and Android Studio.
- Gemini 3.5 Flash Cyber is a security-focused variant built on 3.5 Flash, designed to detect, validate and patch code vulnerabilities at scale at a lower cost per token than larger models.
- Flash Cyber is integrated into Google’s CodeMender tool, which uses multiple Cyber agents to automate security remediation workflows.
- Access to Gemini 3.5 Flash Cyber will be limited at first, available exclusively through CodeMender in a restricted pilot for governments and selected strategic partners.
- The broader Gemini 3.x Flash lineup reflects Google’s effort to segment models by speed, cost and specialization, from everyday tasks (Flash-Lite) to more advanced agentic and security use cases (3.6 Flash and Flash Cyber).
Impact
By expanding the Flash line into 3.6 Flash, 3.5 Flash-Lite and a dedicated Cyber variant, Google DeepMind is sharpening Gemini’s position in agentic AI against rivals like OpenAI and Anthropic. The split between high-throughput, low-cost Flash-Lite and more capable Flash and Cyber models gives enterprises clearer options for everyday workflows versus specialized security automation, while the limited-access Cyber rollout signals a cautious approach to powerful code-auditing capabilities in sensitive, regulated environments.
