Open weights
A model whose trained parameters you can download and run yourself, no API required.
Think of it like
Getting the actual recipe and cooking at home, versus only being able to order the dish.
Example
With an open-weights model you can self-host on your own servers and fine-tune it privately.
How it actually works
Open weights means the parameter files are released, so you can run, inspect, and fine-tune locally. It’s not the same as fully open source — training data and code may stay private, and licenses can restrict use. The trade is control and privacy for the burden of hosting it yourself.
For product teams
Open weights buy you self-hosting, privacy, and customization — at the cost of running the infrastructure.
For engineers
Downloadable parameter files for local inference/fine-tuning; distinct from open training data or code.
Related
- Closed model — The opposite is a closed model.
- Weights — Ships the weights.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome