Authenticate the Opencode CLI with your Google account. This plugin enables you to use your existing Gemini plan and quotas (including the free tier) directly within Opencode, bypassing separate API billing.
- Opencode CLI installed.
- A Google account with access to Gemini.
Add the plugin to your Opencode configuration file
(~/.config/opencode/opencode.json or similar):
{
"$schema": "https://opencode.ai/config.json",
"plugin": ["opencode-gemini-auth@latest"]
}Important
If you're using a paid Gemini Code Assist subscription (Standard/Enterprise),
explicitly configure a Google Cloud projectId. Free tier accounts should
auto-provision a managed project, but you can still set projectId to force
a specific project.
-
Login: Run the authentication command in your terminal:
opencode auth login
-
Select Provider: Choose Google from the list.
-
Authenticate: Select OAuth with Google (Gemini CLI).
- A browser window will open for you to approve the access.
- The plugin spins up a temporary local server to capture the callback.
- If the local server fails (e.g., port in use or headless environment), you can manually paste the callback URL or just the authorization code.
Once authenticated, Opencode will use your Google account for Gemini requests.
By default, the plugin attempts to provision or find a suitable Google Cloud
project. To force a specific project, set the projectId in your configuration
or via environment variables:
File: ~/.config/opencode/opencode.json
{
"provider": {
"google": {
"options": {
"projectId": "your-specific-project-id"
}
}
}
}You can also set OPENCODE_GEMINI_PROJECT_ID, GOOGLE_CLOUD_PROJECT, or
GOOGLE_CLOUD_PROJECT_ID to supply the project ID via environment variables.
Below are example model entries you can add under provider.google.models in your
Opencode config. Each model can include an options.thinkingConfig block to
enable "thinking" features.
{
"provider": {
"google": {
"models": {
"gemini-2.5-flash": {
"options": {
"thinkingConfig": {
"thinkingBudget": 8192,
"includeThoughts": true
}
}
},
"gemini-2.5-pro": {
"options": {
"thinkingConfig": {
"thinkingBudget": 8192,
"includeThoughts": true
}
}
},
"gemini-3-flash-preview": {
"options": {
"thinkingConfig": {
"thinkingLevel": "high",
"includeThoughts": true
}
}
},
"gemini-3-pro-preview": {
"options": {
"thinkingConfig": {
"thinkingLevel": "high",
"includeThoughts": true
}
}
}
}
}
}
}Note: Available model names and previews may change—check Google's documentation or the Gemini product page for the current model identifiers.
The plugin supports configuring Gemini "thinking" features per-model via
thinkingConfig. The available fields depend on the model family:
- For Gemini 3 models: use
thinkingLevelwith values"low"or"high". - For Gemini 2.5 models: use
thinkingBudget(token count). includeThoughts(boolean) controls whether the model emits internal thoughts.
A combined example showing both model types:
{
"provider": {
"google": {
"models": {
"gemini-3-pro-preview": {
"options": {
"thinkingConfig": {
"thinkingLevel": "high",
"includeThoughts": true
}
}
},
"gemini-2.5-flash": {
"options": {
"thinkingConfig": {
"thinkingBudget": 8192,
"includeThoughts": true
}
}
}
}
}
}
}If you don't set a thinkingConfig for a model, the plugin will use default
behavior for that model.
If automatic provisioning fails, you may need to set up the project manually:
- Go to the Google Cloud Console.
- Create or select a project.
- Enable the Gemini for Google Cloud API
(
cloudaicompanion.googleapis.com). - Configure the
projectIdin your Opencode config as shown above.
Common causes of 429 RESOURCE_EXHAUSTED or QUOTA_EXHAUSTED:
- No project ID configured: the plugin uses a managed free-tier project, which has lower quotas.
- Model-specific limits: quotas are tracked per model (e.g.,
gemini-3-pro-previewvsgemini-3-flash-preview). - Large prompts: OAuth/Code Assist does not support cached content, so long system prompts and history can burn quota quickly.
- Parallel sessions: multiple Opencode windows can drain the same bucket.
Notes:
- Gemini CLI auto-fallbacks: the official CLI may fall back to Flash when Pro quotas are exhausted, so it can appear to “work” even if the Pro bucket is depleted.
- Paid plans still require a project: to use paid quotas in Opencode, set
provider.google.options.projectId(orOPENCODE_GEMINI_PROJECT_ID) and re-authenticate.
To view detailed logs of Gemini requests and responses, set the
OPENCODE_GEMINI_DEBUG environment variable:
OPENCODE_GEMINI_DEBUG=1 opencodeThis will generate gemini-debug-<timestamp>.log files in your working
directory containing sanitized request/response details.
This plugin mirrors the official Gemini CLI OAuth flow and Code Assist endpoints. In particular, project onboarding and quota retry handling follow the same behavior patterns as the Gemini CLI.
- Gemini CLI repository: https://github.com/google-gemini/gemini-cli
- Gemini CLI quota documentation: https://developers.google.com/gemini-code-assist/resources/quotas
Opencode does not automatically update plugins. To update to the latest version, you must clear the cached plugin:
# Clear the specific plugin cache
rm -rf ~/.cache/opencode/node_modules/opencode-gemini-auth
# Run Opencode to trigger a fresh install
opencodeTo develop on this plugin locally:
-
Clone:
git clone https://github.com/jenslys/opencode-gemini-auth.git cd opencode-gemini-auth bun install -
Link: Update your Opencode config to point to your local directory using a
file://URL:{ "plugin": ["file:///absolute/path/to/opencode-gemini-auth"] }
MIT