Not for services aimed at under-18s
Generative AI Services may not be used in a site or app directed towards, or likely to be accessed by, individuals under 18. [7]
Guides · ai · Popular APIs
Use Gemini and partner models on Google Cloud: a project with billing, the aiplatform.googleapis.com API enabled, IAM roles, and authentication via Application Default Credentials / service account (production) or an API key (testing, or express mode with no billing).
Researched · not yet run liveLast researched 25 October 2026 · Google Cloud · provider site
A Google account owns or joins a Google Cloud project; the project is linked to a Cloud Billing account and has the Vertex AI (Agent Platform) API enabled. Quotas and spend are per project; Standard PayGo tiers are set by organisation spend.
Google now documents Vertex AI as "Gemini Enterprise Agent Platform"; the API is still aiplatform.googleapis.com.
Steps marked “human step” are ones only you can do: verapli never signs in, submits, pays or accepts terms for you.
Express mode (@gmail.com only) gives an API key with no billing and a 90-day free tier for new users, with low rate limits. Standard setup (project + billing) is needed for production, ADC and higher quotas.
You sign in with a @gmail.com account and accept the terms; an API key is created automatically and is shown on the API Keys page. Upgrade later with "Activate now".
verapli captures GOOGLE_API_KEY here by plain code: encrypted in your browser, never shown to AI.
Pick a project in the console project selector or create one (needs the Project Creator role).
Where: https://console.cloud.google.com/projectselector2/home/dashboard
You confirm billing is enabled on the project, creating a billing account (or Free Trial) and entering payment details yourself if needed.
Where: https://console.cloud.google.com/billing
Enable aiplatform.googleapis.com on the project (needs serviceusage.services.enable, e.g. Owner or Service Usage Admin), or run gcloud services enable aiplatform.googleapis.com.
Where: https://console.cloud.google.com/apis/library/aiplatform.googleapis.com
Give the calling user or service account roles/aiplatform.user (enough for most use); roles/aiplatform.admin for full control.
Where: https://console.cloud.google.com/iam-admin/iam
Locally, install the gcloud CLI, run gcloud init then gcloud auth application-default login. In production, run as a service account with the role above. Set the project ID and location in your client.
Alternatively create an API key for the project from the API Keys page, add restrictions, and keep it out of source control and client-side apps. Google recommends API keys for testing and ADC for production.
verapli captures GOOGLE_API_KEY here by plain code: encrypted in your browser, never shown to AI.
For non-Google models (e.g. Claude), open the model card in Model Garden, click Enable and accept the model's terms yourself. Needs the Consumer Procurement Entitlement Manager role and an org policy that allows partner models.
Plain-English summary, not legal advice. Each card links its source.
Generative AI Services may not be used in a site or app directed towards, or likely to be accessed by, individuals under 18. [7]
You may not use AI/ML outputs to substitute or circumvent a Google model or to create or improve similar models (Google's fine-tuning/distillation features excepted). The Generative AI Prohibited Use Policy applies. [7]
Google will not use Customer Data to train or fine-tune AI/ML models without your prior permission or instruction; generated output is your Customer Data. [7]
Grounded Results and Search Suggestions may not be cached, syndicated or trained on, apart from narrow evaluation and chat-history exceptions. [7]
Anyone with your key can use your project's quota, causing charges and access to tuned models and files. Do not ship keys in mobile or web clients. [2]
| Env var | Secret | Format | What it is |
|---|---|---|---|
GOOGLE_CLOUD_PROJECT | not secret | my-project-… | Google Cloud project ID used for Vertex AI calls |
GOOGLE_CLOUD_LOCATION | not secret | global | Region or global endpoint for requests (global is recommended for Standard PayGo) |
GOOGLE_API_KEY | secret | AIza… | API key for Gemini on Vertex AI (express mode or testing) |
Source [4]
The same entry as JSON: /api/v1/entries/google/vertex-ai · requirements /api/v1/requirements/google/vertex-ai · terms /api/v1/terms/google/vertex-ai.
claude mcp add --transport http verapli https://verapli.com/mcp
> "What do I need for Google Vertex AI (Gemini Enterprise Agent Platform) – project, API and credentials, and how long will it take?"
→ tools/call requirements { "provider_id": "google/vertex-ai" }Provider rules change. This page is generated from verapli registry version 2026.1025.387; spotted something out of date? Tell us. Google Cloud is not affiliated with verapli.