Google has also entered September with a new generation of Gemini models, including Gemini 3.8 Flash, designed to provide a faster and more cost-efficient option for reasoning, multimodal tasks and agentic workflows. The model is aimed at situations where organizations need AI to process large volumes of requests without using the most expensive frontier model for every task.

For managers, this highlights an increasingly important principle: not every task needs the most powerful AI model. A company could use a high-end model for strategic analysis and complex decisions, while a faster model handles routine document processing, summaries, classification, reporting or internal queries. Choosing the right model for each task can therefore become part of managing AI costs.