Gemini Integration

Connecting the Google Gemini API to Starnion lets you use Gemini models in your conversations. Take advantage of Gemini’s powerful features like image analysis and multimodal processing.

Overview

With the Gemini integration you can:

Chat models: Use Gemini 2.0 Flash, Gemini 1.5 Pro, Gemini 1.5 Flash
Embeddings: gemini-embedding-001 (768 dimensions) support
Image analysis: Upload images for content analysis and description
Multimodal processing: Combined text and image understanding

Default provider: Gemini is Starnion’s default provider. You can get started with the free tier without any additional API key setup.

Supported Models

Model	Features	Recommended Use
Gemini 2.0 Flash	Fast responses, latest model	Everyday conversations, quick Q&A
Gemini 1.5 Pro	Long context (up to 1M tokens)	Long document analysis, complex reasoning
Gemini 1.5 Flash	Balanced performance	General-purpose conversations
gemini-embedding-001	768-dimensional embeddings	Document vectorization, similarity search

Prerequisites: Get a Google AI Studio API Key

Step 1: Create an API Key

Go to Google AI Studio.
Log in with your Google account.
Click the Get API key button.
Click Create API key → select or create a project.
Copy the generated key (AIza... format).

Free limits: 15 requests per minute, 1,500 requests per day (as of 2025).

Setup

Register API Key in Web UI

Log in to the Starnion web UI.
Go to Settings → Models tab.
Select the Google Gemini provider.
Paste the copied key (AIza...) in the API Key field.
Check the models you want to use (multiple selection allowed).
Click Save.

The API key is validated automatically upon saving.

Usage

Selecting a Model in Conversations

Select a Gemini model from the model dropdown when starting a conversation, or link a Gemini model to a persona for automatic application.

Image Analysis

You: (attach image) What's in this photo?
Bot: This image shows a sunset over the ocean.
     The warm orange and pink hues reflect beautifully on the water surface.

You: (attach chart image) Analyze this chart
Bot: This bar chart shows monthly sales trends.
     Sales have been steadily increasing since March, reaching a peak in June.

Long Document Analysis

You: (attach PDF) Summarize this report
Bot: (Using Gemini 1.5 Pro)
     Key points of this report:
     1. Q1 2026 revenue up 15% year-over-year
     2. MAU grew 30% with new service launch
     ...

Embedding Configuration

Gemini’s embedding model is used for document search, similarity analysis, and more.

Model: gemini-embedding-001
Dimensions: 768
Use cases: Conversation history vectorization, document search, pattern analysis

Embedding settings can be configured in starnion.yaml or via environment variables.

Comparison with Other Providers

Feature	Gemini	OpenAI	Anthropic
Free tier	Yes (1,500/day)	No	No
Max context	1M tokens	128K tokens	200K tokens
Image analysis	Yes	Yes	Yes
Embeddings	Yes	Yes	No

Troubleshooting

“Gemini API key is invalid”

Verify the API key is in AIza... format.
Check on Google AI Studio that the key is active.

“Request limit exceeded” (429 error)

You have exceeded the free limits (15/min, 1,500/day).
Wait a moment or switch to another provider.

Image analysis not working

Verify you are using Gemini 2.0 Flash or Gemini 1.5 Pro.
Check that the image format is supported (JPEG, PNG, GIF, WebP).

FAQ

Q: Can I use Gemini for free indefinitely? A: Yes, within Google AI Studio’s free limits. Charges may apply if you exceed the limits.

Q: Can I configure Gemini alongside other providers? A: Yes, you can register API keys for multiple providers simultaneously and select models per conversation.

Q: Do I need separate setup for Gemini embeddings? A: Once the Gemini API key is registered, embeddings are automatically available. Fine-tune settings in starnion.yaml.