At a glance
- Availability: Experimental (how to enable).
- Auth: API key.
- Connection: The key from
GEMINI_API_KEYis sent as thex-goog-api-keyheader. - Docs: https://ai.google.dev/gemini-api/docs/api-key
Credentials
Set these per environment. See Connect an integration.| Variable | Required | Description |
|---|---|---|
GEMINI_API_KEY | Yes | Gemini API key from Google AI Studio (starts with AIza) Docs. |
Setup
- Sign in to Google AI Studio: Go to https://aistudio.google.com and sign in with a Google account. The free tier works for testing without billing.
- Create an API key: Open https://aistudio.google.com/app/apikey and click ‘Create API key’. Choose an existing Google Cloud project or let AI Studio create one for you.
- Store the key: Copy the key and add it to your .env file as GEMINI_API_KEY=AIza…
- Verify access: Run the List Models tool to confirm the key works. A 400 with API_KEY_INVALID means the key is wrong or restricted.
- This connector uses the Gemini Developer API key (x-goog-api-key header), not Google OAuth - do not use Google Cloud OAuth credentials here
- Free-tier keys are rate limited; attach a billing-enabled Google Cloud project for higher limits
- Generation and embedding calls are billed per token on paid tiers
Tools
| Tool | Access | Description |
|---|---|---|
| List Models | Read | List the Gemini models available to the API key |
| Get Model | Read | Get details about a specific Gemini model |
| Generate Content | Write | Generate a model response from text or multimodal content |
| Embed Content | Write | Create an embedding vector for input text |
| Count Tokens | Read | Count the tokens a prompt would use for a given model |
| Generate Image | Write | Generate or edit an image with a Nano Banana model via generateContent; the response returns image parts as base64 inlineData |
| Edit Image | Write | Edit an existing image with a Nano Banana model: send the source image as a base64 inlineData part plus a text instruction (add/remove/modify elements, change style, adjust color grading); the edited image is returned as base64 inlineData |
| Generate Image (Imagen) | Write | Generate text-to-image samples with an Imagen 4 model via the predict endpoint; returns base64-encoded images in predictions |
| Upload File | Write | Upload content to the Files API for use in later generateContent calls via its file URI; this simple media upload sends the decoded bytes directly as the request body (a custom displayName requires the multipart or resumable X-Goog-Upload protocol), and files expire after 48 hours |
| List Files | Read | List files uploaded to the Files API for this project |
| Get File | Read | Get metadata for an uploaded file, including its uri and processing state |
| Delete File | Write | Delete an uploaded file before its automatic 48-hour expiry |
Example prompts
- List the Gemini models available to my API key and summarize their token limits.
- Use Gemini to generate a response to a prompt I provide.
- Generate an image with a Nano Banana model from a description I provide.