Glossary
Open-weight model
A model whose trained parameters are published so anyone can download and run it.
An open-weight model is one whose weights, the trained parameters, are released publicly. Others can run the model on their own hardware, study it, and fine-tune it, rather than reaching it only through the developer's app or API.
Open-weight is not the same as open-source. Many open-weight models come with licenses that restrict some uses, and most do not include the training data or full training code.
Supporters say open weights support research, competition, and local control of data. Critics point out that once weights are released, built-in safeguards can be removed and the release cannot be recalled.