Google releases Gemini 3.6 Flash, 3.5 Flash-Lite and 3.5 Flash Cyber

Large language modelsCybersecurityUnited States

Published: 21 July 2026. Google introduced efficiency-focused models for high-volume and agentic workloads. Gemini 3.6 Flash supports a 1,048,576-token input context, function calling, code execution, search grounding and preview computer use. Google says it uses fewer output tokens than 3.5 Flash. Flash Cyber is a specialised vulnerability-finding and patching model offered initially through a restricted pilot.

Why it matters: Model competition is shifting from raw benchmark leadership to cost per completed workflow, tool reliability and domain specialisation. Google announcement · Google DeepMind · a report