Choose a Model

commit-msg-ai works with any model available in Ollama. You need at least one downloaded.

List your models

See what’s already on your machine:

ollama list

If the list is empty, you’ll need to pull one.

Model Size Best for
llama3.2 ~2 GB General purpose, fast, good default
qwen2.5-coder ~4.7 GB Code understanding, best quality
mistral ~2.3 GB Fast and capable
codellama ~3.8 GB Code-focused

Pull a model

# Lightweight and fast (recommended to start)
ollama pull llama3.2

# Better code understanding
ollama pull qwen2.5-coder

Start with llama3.2 if you’re unsure. It’s small, fast, and produces good commit messages. You can always switch later.

Browse all models

The full catalog is at ollama.com/library. Any model that works with Ollama works with commit-msg-ai.

Test your model

Make sure it’s working:

ollama run llama3.2 "Say hello"

If you get a response, you’re ready to install commit-msg-ai.


Next: Install commit-msg-ai