Glossary
Open-weight model
An AI model whose trained weights you can download and run yourself, under the publisher's license.
An open-weight model is an AI model whose trained parameters, called weights, are published so anyone can download, run and often fine-tune them under the publisher's license. It is not the same as open source: the training data and code may not be released, and the license can limit use. Open-weight models let you host the model yourself and keep data inside your own environment.
The Open Source Initiative has published its own definition of open source AI, which sets a higher bar than releasing weights alone. Many popular models are open-weight under custom licenses, so read the license before you build on one.
Choosing one means taking on hosting, security updates and monitoring that a hosted API would otherwise handle. Teams pick open-weight models for data control, predictable cost at volume or the freedom to fine-tune. See [private LLM development](/solutions/private-llm-development).
Start in 60 seconds
Put this to work on your project.
Get an AI-assisted, architect-reviewed scope, cost range, and timeline. Free, no login.