Local AI Setup - Stanza Help Center

📖 Overview

Stanza's Local AI feature allows you to run AI models directly on your Mac, ensuring complete privacy and offline functionality. Using Ollama, you can install and run powerful language models locally without any internet connection or API costs.

🔒 Privacy First: With Local AI, all your text is processed entirely on your device. Nothing is sent to external servers, giving you complete control over your data.

⚙️ Auto Setup Steps

Follow these simple steps to set up Local AI in Stanza:

Open Stanza Settings: Launch Stanza on your Mac and click on the "Settings" button in the app.
Choose Local AI as Service Type: In the Settings window, navigate to the "Service Type" section and select "Local AI" from the available options.
Choose Your Preferred Model: Select from the available models:
Scroll to Local AI Setup Section: Scroll down in the Settings window to find the "Local AI Setup" section.
Click Auto Setup Button: Click the "Auto Setup" button. This will:
Monitor Download Progress: You can see the download progress directly in the Settings window. The model download may take a few minutes depending on your internet connection speed:
Ready to Use: Once the download completes and the model is loaded, you're all set! Stanza will automatically use your local AI model for grammar checking.

✅ That's it! Once setup is complete, you can use Stanza completely offline with your local AI model. No internet connection required, no API costs, and complete privacy.

🤖 About Ollama

Ollama is an open-source tool that makes it easy to run large language models locally on your Mac. It handles all the complex setup, model management, and optimization automatically.

Automatic Installation: Stanza's auto setup installs Ollama for you - no manual configuration needed
Background Service: Ollama runs as a background service on your Mac, ready to process requests whenever Stanza needs it
Model Management: Ollama handles downloading, storing, and running your chosen model efficiently
Resource Efficient: Optimized to use your Mac's resources efficiently while maintaining good performance

💡 Note: Ollama will continue running in the background after setup. You can stop it anytime from Stanza's settings if needed, but it needs to be running for Local AI to work.

📊 Model Comparison

Compare the available local AI models to choose the best one for your needs:

Feature	Llama 3.2 3B	Qwen 3 1.7B
Model Size	3 Billion Parameters	1.7 Billion Parameters
Download Size	~2-3 GB	~1-1.5 GB
RAM Usage	~4-6 GB	~2-3 GB
Speed	Fast	Very Fast
Quality	Excellent	Very Good
Best For	Users with more RAM, need higher quality	Users with limited RAM, need faster responses
Developer	Meta (Facebook)	Alibaba Cloud

Which Model Should You Choose?

Choose Llama 3.2 3B if:

You have 8GB or more RAM available
You prioritize grammar checking quality over speed
You have sufficient storage space (~3GB)
You want the most accurate grammar corrections

Choose Qwen 3 1.7B if:

You have limited RAM (4-6GB available)
You want faster response times
You want to save storage space (~1.5GB)
You need a lightweight solution that still provides excellent results

💡 Tip: Both models provide excellent grammar checking results. The main difference is in resource usage and speed. If you're unsure, start with Qwen 3 1.7B for its efficiency, and you can always switch to Llama 3.2 3B later if you want higher quality.

💻 System Requirements

To use Local AI in Stanza, your Mac needs to meet these requirements:

macOS: macOS 12.0 (Monterey) or later
RAM:
- Minimum 4GB available RAM (for Qwen 3 1.7B)
- Recommended 8GB+ available RAM (for Llama 3.2 3B)
Storage:
- At least 3GB free space for model storage
- Additional space for Ollama service files
Internet Connection:
- Required only for initial setup and model download
- Not needed after setup is complete

⚠️ Important: Make sure you have enough free RAM available. If your Mac is running low on memory, close other applications before using Local AI to ensure optimal performance.

❓ Frequently Asked Questions

Do I need an internet connection to use Local AI?

You only need an internet connection during the initial setup to download the Ollama service and your chosen model. Once setup is complete, Local AI works completely offline - no internet connection required!

Can I switch between models after setup?

Yes! You can switch between Llama 3.2 3B and Qwen 3 1.7B anytime in Stanza's settings. If you switch to a model you haven't downloaded yet, Stanza will automatically download it for you.

How much disk space do the models take?

Llama 3.2 3B: Approximately 2-3 GB
Qwen 3 1.7B: Approximately 1-1.5 GB

Both models are stored locally on your Mac, so make sure you have enough free space.

Will Local AI slow down my Mac?

Local AI models are optimized to run efficiently on your Mac. While they do use RAM and CPU resources, the impact is minimal. If you notice any slowdown, try closing other applications or switching to the lighter Qwen 3 1.7B model.

Can I use Local AI alongside BYOK (Bring Your Own Key)?

Yes! You can switch between Local AI and BYOK anytime in Stanza's settings. Local AI works offline, while BYOK requires an internet connection to use cloud-based models.

What if the auto setup fails?

If auto setup encounters any issues, make sure you have:

Sufficient disk space available
An active internet connection for the initial download
Enough RAM available (close other applications if needed)
Administrator permissions (may be required for Ollama installation)

If problems persist, contact support for assistance.

📧 Need Help?

If you encounter any issues setting up Local AI or have questions about the models, feel free to reach out to us at support@stanza-app.com

Local AI Setup Guide