Skip to content

gemini-setup.md

Latest commit

 

History

History
71 lines (46 loc) · 1.91 KB

gemini-setup.md

File metadata and controls

71 lines (46 loc) · 1.91 KB

Gemini API Setup Guide

Overview

Verily uses Google's Gemini 2.0 Flash model to verify video submissions. This guide walks through setting up the Gemini API for your development environment.

Steps

1. Create a Google Cloud Project

gcloud projects create verily-app --name="Verily"
gcloud config set project verily-app

Or via the GCP Console.

2. Enable the Generative Language API

gcloud services enable generativelanguage.googleapis.com

3. Create an API Key

  1. Go to API Credentials
  2. Click "Create Credentials" > "API Key"
  3. Restrict the key to "Generative Language API" only
  4. Copy the key

4. Configure Serverpod

Add the API key to verily_server/config/passwords.yaml:

development:
  # ... existing keys ...
  geminiApiKey: "YOUR_API_KEY_HERE"

5. Rate Limits & Billing

  • Free tier: 60 requests/minute, 1500 requests/day
  • Paid tier: Higher limits with pay-per-use
  • Set up billing alerts to avoid unexpected charges

Without Gemini API Key

The server gracefully handles a missing Gemini API key:

  • Action browsing, creation, and all non-verification features work normally
  • Video submissions will return a "verification service unavailable" error
  • The submission status will show as error with an appropriate message

Testing

To verify your setup:

  1. Start the server: server:start
  2. Submit a video through the app
  3. Check server logs for Gemini API responses

Model Selection

We use gemini-2.0-flash for:

  • Fast response times (< 5 seconds for most videos)
  • Good accuracy for visual content analysis
  • Cost-effective for high-volume verification

For higher accuracy needs, you can change the model in gemini_service.dart to gemini-2.0-pro, though this will increase latency and cost.