🤖 AI Provider Setup Guide

Configure AI providers for advanced grammar checking and tone rewriting.

📋 Overview

OpenGrammar supports 6 AI providers, giving you flexibility in cost, quality, and privacy.

🎯 Quick Setup

Step 1: Choose Your Provider

For Free Usage:

• Provider: Groq
• Free tier: 100 requests/day
• Quality: Excellent (Llama 3.1 70B)

For Best Quality:

• Provider: OpenAI
• Model: GPT-4o-mini
• Cost: ~$0.15 per 1K requests

For Privacy:

• Provider: Ollama (Local)
• 100% offline
• Free (your hardware)

Step 2: Get API Key

Follow the guide for your chosen provider below.

Step 3: Configure Extension

1. Click OpenGrammar icon
2. Click Settings (gear icon)
3. Select Provider from dropdown
4. Enter API Key
5. Select Model
6. Click Save

⚡ Groq (Recommended - Free)

Overview

• Free Tier: 100 requests/day
• Speed: Fastest (LPU technology)
• Quality: Excellent (Llama 3.1 70B)
• Setup Time: 2 minutes

Step 1: Create Account

1. Visit Groq Console
2. Click Sign Up
3. Complete registration

Step 2: Get API Key

1. Go to API Keys in left sidebar
2. Click Create API Key
3. Give it a name (e.g., "OpenGrammar")
4. Copy the key (starts with gsk_)
5. ⚠️ Save it now - you can't see it again!

Step 3: Configure in Extension

1. Click OpenGrammar icon → Settings
2. Provider: Groq
3. API Key: gsk_xxx (paste your key)
4. Model: llama-3.1-70b-versatile
5. Click Save

Available Models

Testing

# Test Groq API
curl https://api.groq.com/openai/v1/chat/completions \
  -H "Authorization: Bearer gsk_xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "llama-3.1-70b-versatile",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Troubleshooting

• Error: Invalid API key - Check key starts with gsk_
• Error: Rate limit - Free tier is 100/day, wait 24 hours
• Error: Model not found - Use exact model name

🟢 OpenAI (Best Quality)

Overview

• Quality: Best available
• Speed: Good
• Cost: $0.15 per 1K requests (GPT-4o-mini)
• Setup Time: 5 minutes

Step 1: Create Account

1. Visit OpenAI Platform
2. Click Sign Up
3. Complete registration
4. Add phone number (required)

Step 2: Add Credits

1. Go to Settings → Billing
2. Click Add payment method
3. Add minimum $5 credit (recommended)

Step 3: Get API Key

1. Go to API Keys section
2. Click Create new secret key
3. Give it a name (e.g., "OpenGrammar")
4. Copy the key (starts with sk-)
5. ⚠️ Save it now - can't see it again!

Step 4: Configure in Extension

1. Click OpenGrammar icon → Settings
2. Provider: OpenAI
3. API Key: sk-xxx (paste your key)
4. Model: gpt-4o-mini (recommended)
5. Click Save

Available Models

Cost Estimates

• Light user (50 checks/day): ~$0.25/month
• Medium user (200 checks/day): ~$1/month
• Heavy user (1000 checks/day): ~$5/month

Testing

# Test OpenAI API
curl https://api.openai.com/v1/chat/completions \
  -H "Authorization: Bearer sk-xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o-mini",
    "messages": [{"role": "user", "content": "Fix: me and him went"}]
  }'

🌐 OpenRouter (100+ Models)

Overview

• Models: 100+ (Claude, Llama, Mistral, etc.)
• Cost: Pay-per-use, varies by model
• Speed: Good
• Setup Time: 5 minutes

Step 1: Create Account

1. Visit OpenRouter
2. Click Sign In
3. Sign in with Google or GitHub

Step 2: Get API Key

1. Go to Keys section
2. Click Create Key
3. Give it a name
4. Copy the key

Step 3: Add Credits

1. Go to Credits section
2. Add minimum $5 credit
3. Choose payment method

Step 4: Configure in Extension

1. Click OpenGrammar icon → Settings
2. Provider: OpenRouter
3. API Key: (paste your key)
4. Model: anthropic/claude-3.5-sonnet (recommended)
5. Click Save

Popular Models

Testing

# Test OpenRouter API
curl https://openrouter.ai/api/v1/chat/completions \
  -H "Authorization: Bearer xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "anthropic/claude-3.5-sonnet",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

🔷 Together AI (Open-Source)

Overview

• Models: Open-source (Llama, Mistral, etc.)
• Cost: Very cheap ($0.01-0.10 per 1K)
• Speed: Fast
• Setup Time: 5 minutes

Step 1: Create Account

1. Visit Together AI
2. Click Sign Up
3. Complete registration

Step 2: Get API Key

1. Go to Settings → API Keys
2. Click Create API Key
3. Copy the key

Step 3: Add Credits

1. Go to Billing section
2. Add credits (minimum $5)
3. Get $25 free credit on signup

Step 4: Configure in Extension

1. Click OpenGrammar icon → Settings
2. Provider: Together AI
3. API Key: (paste your key)
4. Model: meta-llama/Meta-Llama-3.1-70B-Instruct
5. Click Save

Available Models

🦙 Ollama (Local - Offline)

Overview

• Privacy: 100% local, offline
• Cost: Free (your hardware)
• Speed: Depends on hardware
• Setup Time: 15 minutes

Prerequisites

• 4GB+ RAM (for smaller models)
• 8GB+ RAM (for larger models)
• NVIDIA GPU (optional, for acceleration)

Step 1: Install Ollama

Linux:

curl -fsSL https://ollama.com/install.sh | sh

macOS:

brew install ollama

Windows:

1. Download from ollama.com
2. Run installer

Step 2: Start Ollama Service

# Start Ollama (runs in background)
ollama serve

Step 3: Pull Models

# Small & fast (recommended for grammar)
ollama pull qwen2.5:0.5b      # 400MB, very fast
ollama pull qwen2.5:1.5b      # 1GB, balanced

# Medium quality
ollama pull phi4-mini:3.8b    # 2.5GB, great quality
ollama pull llama3.2:3b       # 2GB, good balance

# High quality (requires GPU)
ollama pull llama3.2:7b       # 4GB, requires 8GB+ RAM
ollama pull mistral:7b        # 4GB, requires 8GB+ RAM

Step 4: Test Ollama

# Test the model
ollama run qwen2.5:1.5b "Fix this: me and him went to store"

Expected output:

"He and I went to the store."

Step 5: Configure in Extension

1. Click OpenGrammar icon → Settings
2. Provider: Ollama (Local)
3. Base URL: http://localhost:11434/v1
4. Model: qwen2.5:1.5b (or your chosen model)
5. API Key: (leave empty - not needed)
6. Click Save

Model Recommendations

GPU Acceleration (NVIDIA)

# Install NVIDIA Container Toolkit
sudo apt-get install -y nvidia-container-toolkit

# Configure Docker
sudo nvidia-ctk runtime configure --runtime=docker
sudo systemctl restart docker

# Ollama will automatically use GPU
ollama run llama3.2:7b

Troubleshooting

• Error: Connection refused - Make sure Ollama is running: ollama serve
• Error: Model not found - Pull the model: ollama pull qwen2.5:1.5b
• Slow performance - Use smaller model or enable GPU

🔧 Custom Provider

Overview

Use your own API endpoint or any OpenAI-compatible API.

Configuration

1. Click OpenGrammar icon → Settings
2. Provider: Custom
3. Base URL: https://your-api.com/v1
4. API Key: (your API key)
5. Model: (your model name)
6. Click Save

Example: LocalAI

# Run LocalAI with Docker
docker run -p 8080:8080 localai/localai

# Configure in extension
Provider: Custom
Base URL: http://localhost:8080/v1
Model: llama-3.1-70b

Example: LM Studio

# LM Studio runs locally
# Configure in extension
Provider: Custom
Base URL: http://localhost:1234/v1
Model: local-model

📊 Provider Comparison

Speed Comparison

Groq (Llama-70B)        ████████████████████ 200ms
OpenAI (GPT-4o-mini)    ██████████ 1s
Together (Llama-70B)    ████████████ 800ms
OpenRouter (Claude)     ██████████████████ 2s
Ollama (Qwen-1.5B)      ██████████████ 500ms (local)

Quality Comparison

OpenAI (GPT-4o)         ████████████████████ 96%
OpenRouter (Claude)     ████████████████████ 96%
OpenAI (GPT-4o-mini)    ██████████████████ 95%
Groq (Llama-70B)        ████████████████ 92%
Ollama (Phi-4-mini)     ██████████████ 88%
Ollama (Qwen-1.5B)      ████████████ 82%

Cost Comparison (per 1K requests)

Ollama (Local)          $0.00 (free)
Groq (Free tier)        $0.00 (100/day free)
Together (Llama-70B)    $0.40
OpenAI (GPT-4o-mini)    $0.15
OpenRouter (Claude)     $3.00

🎯 Provider Selection Guide

Choose Groq If:

• ✅ You want free usage
• ✅ You need fast responses
• ✅ You're okay with 100 requests/day limit
• ✅ You want good quality

Choose OpenAI If:

• ✅ You want the best quality
• ✅ You don't mind paying
• ✅ You need reliability
• ✅ You want GPT-4 level intelligence

Choose OpenRouter If:

• ✅ You want model variety
• ✅ You want to try Claude
• ✅ You want flexibility
• ✅ You need specific models

Choose Together AI If:

• ✅ You want cheap open-source
• ✅ You like Llama models
• ✅ You want good value
• ✅ You need fast inference

Choose Ollama If:

• ✅ Privacy is your priority
• ✅ You have good hardware
• ✅ You want offline access
• ✅ You don't want API costs

🔄 Switching Providers

You can switch providers anytime:

1. Click OpenGrammar icon
2. Click Settings
3. Change Provider dropdown
4. Enter new API key (if required)
5. Select new Model
6. Click Save

Your settings are saved automatically.

📚 Related Documentation

• Backend Deployment - Deploy your backend
• Docker Self-Hosting - Run locally
• Troubleshooting - Common issues

Your AI provider is now configured! Start writing better! ✨