Dictionary term
Open-Weight Model
Also called open weights · open-weight · open source model · open model
An open-weight model is one whose trained weights are published for anyone to download, run, fine-tune and self-host, under a license that varies by release. The training data and training code are usually not released, so most open-weight models do not meet the Open Source Initiative's Open Source AI Definition, which also requires data information and code.
In plain English
An open-weight model is an AI you can download and run on your own computers. You get the finished model, but usually not the data or code used to build it.
Analogy
Like being handed a finished cake you can eat, slice and decorate however you like, without the recipe or the list of ingredients that produced it.
Why it matters
Open-weight models let you run AI on your own hardware for privacy, cost control at volume, or offline use. Read the license before shipping: some are Apache 2.0 or MIT, while others are custom licenses with conditions such as attribution or limits above a revenue or user threshold.
In practice
A clinic that cannot send patient notes to a cloud API runs an open-weight model such as OpenAI's gpt-oss on its own server through Ollama, and checks the Apache 2.0 license before deploying it.
Related terms
Sources
Browse all terms →Last verified