Connect Gemini on Vertex AI

The gemini preset registers Gemini 2.5 Pro and Gemini 2.5 Flash on Google Cloud's Vertex AI. It takes no API key: calls use your Google Cloud credentials, and Vertex AI runs and bills them in a project you name.

1. Log in to Google Cloud

gcloud auth application-default login

That writes your Application Default Credentials. kindgi dev hands them to the runtime it starts, read-only: the file GOOGLE_APPLICATION_CREDENTIALS names, if it's set, else the one this command writes (~/.config/gcloud/application_default_credentials.json). The account needs access to Vertex AI in the project.

2. Register the preset

kindgi providers register --preset=gemini --project=<gcp-project>

--project is the Google Cloud project Vertex AI runs and bills in; the preset refuses to register without it:

Error: preset "gemini" needs --project=<…> (The Google Cloud project Vertex AI bills and authorises against.)

--models registers only some of the models:

kindgi providers register --preset=gemini --project=<gcp-project> --models=gemini-2.5-flash

{
  "providerId": "gemini"
}
✓ Registered gemini: gemini-2.5-flash

What it registers

kindgi providers get gemini

The provider's id is gemini and its region global. Both models list tool-use as their only feature:

Model	Context window	Per 1K input / output tokens
gemini-2.5-pro	1,048,576	$0.00125 / $0.01
gemini-2.5-flash	1,048,576	$0.0003 / $0.0025

An agent that needs structured-output or long-context doesn't route to them. To send an agent to Gemini when other providers are registered too, set preferredProvider: 'gemini' (preferred_provider="gemini" in Python), or require it: see Choose the model an agent uses.