
Google Cloud AI
Managed models and custom AI deployment using Google Cloud accounts and infrastructure.
- Gemini and other models through managed cloud services
- Custom model hosting on Google Cloud infrastructure
- Cloud projects, permissions, regions, and billing
Google Cloud provides managed model APIs alongside infrastructure for hosting custom models. Its AI platform supports application development, model deployment, retrieval, and monitoring within a Cloud project.
Choosing a deployment
- Use managed Gemini and other model endpoints for hosted inference.
- Deploy a supported model through the cloud model catalogue.
- Run a custom model on services such as Kubernetes Engine, Cloud Run, or Compute Engine.
The right option depends on model requirements, operational control, and available regional capacity.
Gemini Developer API or Google Cloud?
Google AI Studio and the Gemini Developer API offer a direct route for experimenting with Gemini and creating an API key. Google Cloud adds cloud-resource management, identity controls, and deployment choices. Google DeepMind is the research organization behind Gemini; it is not a separate application-hosting service.
How pricing works
Managed models use model-specific input, output, caching, and tool rates. Custom model hosting adds compute costs, and retrieval, storage, or application hosting can add separate charges. Use the rate sheet for the selected service and region.
Sources
From the archive
- Gemini 3.8 Flash launches alongside a restricted Cyber variant 2026-09-02
- Claude 4 launches as Claude Code becomes generally available 2025-05-22
- Gemini 2.5 Pro adds reasoning to a long-context model 2025-03-25
- Gemini 2.0 Flash launches as an experimental model 2024-12-11
- Claude 3.5 Sonnet introduces Artifacts alongside the conversation 2024-06-20
- Google releases Gemma 2B and 7B model weights 2024-02-21