Google offers Gemini two ways. The Gemini API uses an API key from Google AI Studio, and Google says most developers should use it unless they need specific enterprise controls. Vertex AI, which Google's documentation now calls Gemini Enterprise Agent Platform, runs in your Google Cloud project. It uses service accounts and IAM, and adds enterprise terms for data processing, support with SLAs, compliance certifications such as HIPAA and SOC 2, and Provisioned Throughput for dedicated capacity. Both use the same Google Gen AI SDK; the client is set up with a project and location for Vertex AI. Supported regions may differ between the two, and models tuned in AI Studio must be retrained on the platform. We choose by who owns the data, where it must stay and what support you need.