Member-only story
Open-Weight vs. Closed-Weight AI: Who Really Controls the Model?
The weights are public. The secrets aren’t.
Download the model. Run it on your own machine. Fine-tune it on your own data.
It sounds like open source. In most cases, it isn’t.
Not a Medium member? Read it free here.
Every week, a new AI model gets called “open source.” Kimi K3. DeepSeek. Qwen. Llama. gpt-oss.
Almost none of them meet the formal definition. They are open-weight models, and the difference decides what you can build, what you can trust, and what can be taken away from you.
Open weights give you the finished model. Open source gives you everything needed to rebuild it.
This article explains what model weights are, how open-weight and closed-weight models differ, and why the two terms keep getting mixed up.
What Are Model Weights?
Think of an AI model as a giant machine with billions of dials.
Training is the process of turning those dials, one tiny adjustment at a time, until the…
