Local AI Engineering with Ollama

Sharing Your Model

There are mainly 2 ways to share your model. The first is obvious: share your Modelfile (e.g., in a Git repo) and anyone who wants to get the commitwriter model should rebuild it:

# Clone the repo
# git clone..

ollama create commitwriter -f Modelfile

This is the fastest way to share within a team. However, it only works if they can resolve your FROM line. Since yours points at granite4.1:3b, they need that base pulled (Ollama pulls it automatically on create if it is missing). If your FROM had pointed at a local .gguf file instead, you would have to ship that file too.

The second way is to share your model on ollama.com:

# Register your machine by singing in
ollama signin

# Rename the model into your namespace (your username)
ollama cp commitwriter \
    $OLLAMA_USER/commitwriter

# Push it
ollama push $OLLAMA_USER/commitwriter