Open-weights model: An open-weights model is an AI model whose trained parameters (weights) are published for anyone to download, run, and modify, usually under a license that sets limits on commercial use or redistribution.
When a lab releases a model as open weights, it publishes the file of numbers the model learned during training. You can download that file and run the model on your own hardware or a cloud server you control, with no vendor in the middle. Meta's Llama family, Mistral's models, Google's Gemma, Alibaba's Qwen, DeepSeek's models, and OpenAI's gpt-oss models are well-known examples.
Open weights is not the same as open source. Most releases include the weights and enough code to run the model, but not the training data or the full training recipe. Licenses vary too: some are permissive, others restrict certain uses or require a separate agreement above a user threshold. Read the license before you build a product on one.
The practical tradeoff is control versus convenience. Running your own model means your data never leaves your environment and your costs are predictable, but you own the setup, the security, the updates, and the hardware bill. Hosted proprietary models are easier to start with and usually stronger at the top end, though the gap narrows every year.
Example at work
A healthcare billing company wants to draft appeal letters from claim data but cannot send patient records to a consumer chatbot. Its IT team runs an open-weights model on a private cloud server, so the PHI stays inside the company's own HIPAA-covered environment and the compliance officer signs off.
Why it matters
If your work involves confidential data, open-weights models are often the path your security team will approve first, because nothing leaves your network. They also give you a fallback when a vendor raises prices or changes a model you depend on. You do not need to run one yourself to benefit; many hosting providers offer them by API.
Related terms
- Proprietary modelA proprietary model is an AI model whose weights are kept private by the company that built it. You use it through the company's app or API, and you cannot download, inspect, or self-host it.
- Model weightsModel weights are the numbers inside a trained neural network that determine how it responds to input. They are what training produces and what a company ships or protects; releasing them publicly is what makes a model 'open weights.'
- ParametersParameters are the adjustable numbers inside an AI model, mainly its weights, that are learned during training. The parameter count, from a few billion to over a trillion, is the standard shorthand for a model's size and rough capability.
- Fine-tuningFine-tuning is extra training that adjusts an existing AI model's weights using a smaller, targeted set of examples so it performs better on a specific task, style, or domain. It changes the model itself, unlike prompting, which only changes the instructions.
- Data privacyData privacy, in the context of AI, is the set of rules and practices that govern what information you put into an AI system, who can see it, how long the vendor keeps it, and whether it is used to train future models.