Ollama Setup Guide

What is Ollama?

Ollama runs AI models entirely on your computer. No internet required. No API keys. No data leaves your machine. It's the default AI provider for DevDiff.

Step-by-Step Setup

1. Install Ollama

Platform	Instructions
Windows	Download installer → Run OllamaSetup.exe
macOS	brew install ollama or download .dmg
Linux	curl -fsSL https://ollama.com/install.sh | sh

2. Pull a Model

# Recommended: llama3.2 3B (2GB, works on most machines)
ollama pull llama3.2:3b

# Better quality (4GB, needs 8GB RAM)
ollama pull llama3.1:8b

# Code specialist (7GB, needs 16GB RAM)
ollama pull codellama:13b

3. Verify

# Check installed models
ollama list

# Test the model
ollama run llama3.2:3b "Explain what git diff does in one sentence"

# Should respond with something like:
# "Git diff shows the changes between commits, branches, or the working directory..."

4. Configure DevDiff

DevDiff auto-detects Ollama. If it's running on localhost:11434, no config needed.

// .devdiff.config.js — explicit config (optional)
export default {
  ai: {
    providers: [
      {
        name: "local-ollama",
        url: "ollama://llama3.2:3b", // Model to use
        priority: 1, // Try first
      },
    ],
  },
};

5. Generate Changelog

devdiff generate

---

Which Model Should I Choose?

┌──────────────────────────────────────────────────────────────┐
│  Start here: ollama pull llama3.2:3b                         │
│                                                              │
│  Need better quality? → ollama pull llama3.1:8b              │
│  Doing security reviews? → ollama pull codellama:13b         │
│  Low on RAM (<8GB)? → ollama pull llama3.2:1b                │
└──────────────────────────────────────────────────────────────┘

---

Common Questions

Q: Do I need internet after downloading the model? A: No. Once the model is pulled, everything works offline.

Q: How much disk space do I need? A: 2GB for llama3.2:3b, 4GB for llama3.1:8b, 7GB for codellama:13b.

Q: Can I use multiple models? A: Yes. DevDiff can route to different models based on complexity.

Q: Is it really free? A: Yes. Ollama is open-source. Models are free. No API costs.