When you have thirteen products touching language models, every call costs. Metered API keys are convenient to start, but at scale the spend spikes and becomes hard to control across projects.
Why Vertex over Firebase Blaze
Consolidating AI on Vertex AI inside a Blaze project gives me unified billing, Gemini's Flash models at an aggressive cost, and the option of grounding with Google Search without setting up another integration. I reserve my paid API keys for what truly needs them.