Google’s developer AI offering includes the Gemini API, Google AI Studio for exploring models and developing prompts, and the Gemma family of open models. Its developer site also documents tools for running AI on devices. This profile focuses on that developer offering, rather than all of Google’s AI products.
When does a hosted API make sense, and when is control over deployment worth the extra work?
Google is interesting because using an API and running an open model are different choices about where an application depends on a provider. A company name alone is too broad to explain the tradeoff. Start with the product and the place it runs.