The new Gemini 3.6 Flash model cuts token usage by up to 65% for complex coding tasks while maintaining quality at the same cost. Gemini 3.5 Flash-Lite achieves speeds of 350 output tokens per second, offering a fast and cost-effective option for everyday tasks including agentic search. Additionally, Gemini 3.5 Flash Cyber is introduced as a specialized cybersecurity model for vulnerability detection and patching.