A developer is beginning to explore generative AI capabilities for a new project. They want to quickly experiment with prompting Google's latest foundation models, like Gemini, without needing to set up a full cloud environment or incur significant costs initially. Their primary goal is rapid prototyping and understanding the model's behavior with different inputs. Which Google Cloud tool would be most appropriate for this initial, cost-effective experimentation and prototyping phase?