Gemini API Setup

Overview

Tag-AI can use Google's Gemini API for enhanced cloud-based image tagging. This offers improved tag quality and quantity but requires an API key and internet connectivity.

The Gemini API offers both free and paid tiers, with different rate limits and capabilities.

Getting a Gemini API Key

Step 1: Create a Google AI Studio Account

1. Visit Google AI Studio
2. Sign in with your Google account
3. Accept the terms of service

Step 2: Create an API Key

1. In Google AI Studio, click on the "Get API key" button
2. If you already have a project, select it; otherwise, create a new one
3. Click "Create API key"
4. Your new API key will be displayed - copy it immediately

Step 3: Enable Billing (Optional - For Paid Tier)

If you want to use the paid tier with higher rate limits:

1. Go to the Google Cloud Console
2. Select your project
3. Go to "Billing"
4. Set up a billing account and link it to your project
5. Return to Google AI Studio and verify your paid tier is active

Configuring Tag-AI

Step 1: Open Configuration Editor

1. In Tag-AI, click "Actions" in the menu bar
2. Select "Edit Config"

Step 2: Enter API Key

1. Locate the [tagger_gemini] section
2. Paste your API key into the gemini_api_key field
3. If you're using the paid tier, set paid_version to true

Step 3: Switch to Gemini Processing

1. In the configuration editor, scroll to the [script_to_run] section
2. Change llm_script from local to gemini
3. Click "Save" to apply your changes

Step 4: Test the Connection

1. Return to the main interface
2. Click "Scan" to open the scan window
3. Start a scan to process a few images
4. Check for successful API responses in the output log

Free vs. Paid Tiers

Feature	Free Tier	Paid Tier
Cost	Free	Pay-as-you-go based on usage
Rate Limit	15 requests per minute	Up to 2000 requests per minute
Monthly Quota	Limited (subject to Google's policies)	Based on your billing setup
API Availability	Subject to usage restrictions	Higher availability, SLA options
Image Resolution	Same	Same
Model Quality	Same	Same

Rate Limits

Free Tier Limitations

Tag-AI automatically respects the free tier rate limits by:

• Limiting requests to 15 per minute
• Queuing image processing jobs
• Providing progress indicators during processing

Paid Tier Benefits

With the paid tier:

• Process up to 2000 requests per minute
• Tag large image libraries much faster
• No monthly quotas (pay for what you use)

Rate Limit Configuration

Tag-AI automatically detects your tier based on the paid_version setting:

• paid_version = false: Enforces 15 RPM limit
• paid_version = true: Removes rate limiting (subject to your Google Cloud quota)

Troubleshooting

API Key Issues

If you receive API key errors:

• Verify the key is correctly copied without extra spaces
• Check if the key is still valid in Google AI Studio
• Try generating a new key if problems persist
• Ensure you're using a Gemini API key (not a general Google Cloud API key)

Quota Exceeded

If you see "quota exceeded" errors:

• For free tier: Wait until your quota resets (typically hourly)
• For paid tier: Check your Google Cloud Console for any billing issues
• Consider switching to local LLaVA processing temporarily

Network Issues

If Tag-AI can't connect to the Gemini API:

• Check your internet connection
• Verify no firewall is blocking the connection
• Ensure proxy settings are configured correctly if you use a proxy

Processing Errors

If images fail to process with Gemini:

• Check if the image format is supported
• Verify the image isn't corrupt or too large
• Try reducing the max_image_size value in configuration
• Check the scan window output for specific error messages

Privacy Considerations

Data Transmission

When using the Gemini API:

• Your images are transmitted to Google's servers for processing
• Images are resized before transmission (based on max_image_size setting)
• Transmission occurs over encrypted HTTPS connections

Google's Data Usage

According to Google's terms:

• Google may use your data to improve their services
• Your content remains your property
• Google doesn't claim ownership of your images

Sensitive Data

Consider these guidelines for sensitive images:

• Do not process confidential documents
• Avoid sending images with personal identification information
• Use local processing for medical, legal, or private images
• Review Google's current API terms for up-to-date policies