Google ships 3 new Flash models while flagship 3.5 Pro stays lost in training
google released gemini 3.6 flash, 3.5 flash-lite, and a specialized cyber model, but its frontier 3.5 pro remains delayed. this leaves google competing on efficiency rather than capability as rivals like openai advance.
google expanded its gemini lineup with 3.6 flash, 3.5 flash-lite, and a restricted 3.5 flash cyber model, prioritizing efficiency and specialized use cases. 3.6 flash offers a 1m token context window and reduces costs by up to 65% on specific benchmarks like deepswe, while flash-lite targets low-latency workloads at $2.50 per million output tokens. despite these efficiency gains and the start of gemini 4 training, the delayed 3.5 pro model signals google's struggle to match the capability releases of openai and anthropic. the company is consequently leaning into cost and speed to maintain developer interest.