How to Use Local AI Models with Shakespeare

Want complete control over your AI-powered website building? Local AI models offer enhanced privacy, faster responses, and dramatically reduced costs in your development workflow. With Shakespeare's support for local models including GPT-OSS, DeepSeek-R1, Gemma 3 and other models running on your machine, you can build websites entirely offline while keeping your projects and data completely private.

Why Choose Local AI Models?

Complete Privacy

No data ever leaves your machine
Perfect for sensitive business projects
Full control over your intellectual property
No concerns about terms of service changes

Zero Ongoing Costs

Pay once for hardware, use forever
No per-request charges
No monthly subscriptions
Unlimited usage without budget worries

Full Customization

Fine-tune models for your specific needs
Complete control over model behavior
No rate limits or usage restrictions
Experiment freely without costs

Getting Started: Installing Ollama

System Requirements

| Level | Specs | |-------|-------| | Minimum | 8GB RAM, modern CPU | | Recommended | 16GB+ RAM, GPU with 8GB+ VRAM | | Optimal | 32GB+ RAM, high-end GPU |

Installation Steps

macOS Installation

# Install via Homebrew
brew install ollama

# Or download from ollama.ai
curl -fsSL https://ollama.ai/install.sh | sh

Linux Installation

# Install via curl
curl -fsSL https://ollama.ai/install.sh | sh

# Or use package manager
sudo apt install ollama  # Ubuntu/Debian
sudo pacman -S ollama     # Arch Linux

Windows Installation

Download the installer from ollama.ai
Run the installer as administrator
Restart your system after installation

Choosing the Right Model

For Website Building: Top Local Model Recommendations

GPT-OSS - OpenAI's Open-Weight Powerhouse

ollama pull gpt-oss

Best for: Powerful reasoning, agentic tasks, versatile developer use cases

Trade-offs:

✓ Excellent at complex web development
⚠ Requires decent hardware

DeepSeek-R1 - Enterprise-Grade Reasoning

ollama pull deepseek-r1

Best for: Open reasoning with performance approaching O3 and Gemini 2.5 Pro

Trade-offs:

✓ Leading-edge reasoning capabilities
⚠ Slower inference due to reasoning depth

Gemma 3 - Google's Single-GPU Solution

ollama pull gemma3

Best for: Google's most capable model that runs on a single GPU

Trade-offs:

✓ Excellent performance on consumer hardware
⚠ May need more guidance for complex tasks

Quick Model Comparison

| Model | Size | RAM Required | Speed | Code Quality | Content Quality | |-------|------|--------------|-------|--------------|-----------------| | GPT-OSS | 8GB | 16GB | Fast | Excellent | Excellent | | DeepSeek-R1 | 7GB | 16GB | Medium | Excellent | Very Good | | Gemma 3 | 4GB | 8GB | Very Fast | Very Good | Good |

Configuring CORS for Browser Access

Important: CORS Configuration Required

Since Shakespeare runs in your browser, you need to configure Ollama to accept cross-origin requests (CORS). This is a crucial step that allows Shakespeare to communicate with your local Ollama instance.

Checking CORS Status

First, verify if CORS is already enabled:

curl -X OPTIONS http://localhost:11434 -H "Origin: http://example.com" -H "Access-Control-Request-Method: GET" -I

If you see HTTP/1.1 403 Forbidden, CORS is not enabled and needs configuration.

Enabling CORS by Platform

Enabling CORS on macOS

# Allow all origins (easiest for local development)
launchctl setenv OLLAMA_ORIGINS "*"

# Or specify specific origins for better security
launchctl setenv OLLAMA_ORIGINS "localhost:3000,localhost:5173,shakespeare.app"

# Optional: Make Ollama accessible on your network
launchctl setenv OLLAMA_HOST "0.0.0.0"

# Restart Ollama for changes to take effect

Enabling CORS on Linux

Edit the Ollama service configuration:

sudo systemctl edit ollama.service

Add these environment variables:

[Service]
Environment="OLLAMA_HOST=0.0.0.0"
Environment="OLLAMA_ORIGINS=*"

Then restart the service:

sudo service ollama restart

Enabling CORS on Windows

Open System Properties → Environment Variables
Add new system variables:
- OLLAMA_ORIGINS with value * (or specific origins)
- OLLAMA_HOST with value 0.0.0.0 (optional, for network access)
Restart Ollama from the system tray

Verifying CORS Configuration

After configuration, test again:

curl -X OPTIONS http://localhost:11434 -H "Origin: http://example.com" -H "Access-Control-Request-Method: GET" -I

Success looks like:

HTTP/1.1 204 No Content
Access-Control-Allow-Origin: *

Configuring Local Models in Shakespeare

Step 1: Start Ollama Service

# Start Ollama server (with CORS already configured)
ollama serve

# The service will run on http://localhost:11434

Step 2: Test Your Model

ollama run gpt-oss "Write a simple HTML page with a header"

Step 3: Configure in Shakespeare

Open Shakespeare and go to Settings > AI Settings
Scroll to "Add Custom Provider"
Click to expand the custom provider section
Enter the following configuration:
- Provider Name: "ollama"
- API Endpoint: http://localhost:11434/v1

The Bottom Line

Local Models with Shakespeare Provide

✓ Enhanced privacy - Your code never leaves your machine
✓ Faster responses - No network latency
✓ Reduced costs - Zero API fees after initial setup
✓ Complete control - Run any compatible model

Ready to supercharge your development workflow? Start with GPT-OSS and experience the benefits of local AI models in Shakespeare.

Ready to Build with Shakespeare?

Start building amazing projects with AI-powered development on Nostr.

Turn your ideas into reality through natural conversation with AI.

This article was originally published on Soapbox.pub