Skip to main content
Distributing Modelfiles for every update can be cumbersome. By uploading custom models to the Ollama Model Registry, your team can pull the latest version directly—just like using Docker Hub for container images. In this tutorial, we’ll cover how to:
  1. Set up your Ollama account and trust relationship
  2. Tag and copy your local model
  3. Push the tagged model to the registry
  4. Pull and verify your custom model

​
1. Set Up Your Ollama Account and Trust Relationship

  1. Create an Ollama account at https://ollama.com/ and sign in.
  2. Go to Settings → Ollama Key.
  3. Locate your public SSH key on your local machine:
  1. Display and copy your public key:
  1. Paste the key into your Ollama Ollama Key field and save.
Once added, Ollama will recognize and authenticate your machine for publishing.

​
2. Tag and Copy Your Local Model

First, list your existing local models:
Example output:
Next, create a tagged copy under your Ollama username:
Replace your_username with your actual Ollama account name. Verify the new entry:
Expected result:

​
3. Push Your Model to the Registry

Push the tagged model to make it available on Ollama:
Sample output:
Visit the URL to view details such as architecture, parameter count, quantization, and any custom system instructions.
The image shows a user interface for a model named "harris" with details about its architecture, parameters, and quantization. It includes sections for model information, parameters, system description, template, and license details.

​
4. Pull and Verify the Custom Model

Anyone with access can now pull and run your personalized model:

​
Quick Reference: Ollama CLI Commands

​
Conclusion

Uploading custom models to the Ollama Model Registry simplifies sharing, version control, and collaboration. Experiment with tagging, pushing, and pulling your own models to integrate this workflow into your development process.

Watch Video