Open models, kept by everyone.

Common Weights keeps every model on your machine in one place, and shares it with a worldwide network of everyone who runs it.

• No wasted disk space. Ollama, llama.cpp, vLLM and transformers share one copy of each model.
• Faster, and always there. Models come from whoever has them nearby, a colleague or the next node in your cluster, and keep downloading when the hub is down.
• Know what you run. Every file is checked against what was originally released.

No company, court or government should be able to make an open model disappear. We'll make sure they can't.

Coming soon.