Open-Weights Model
Also known as · open source model · open weights · open model
A model whose trained parameters are publicly released to download and run.
An open-weights model is one whose trained parameters are released publicly, so anyone can download, run, inspect, and fine-tune it on their own hardware. Llama, Mistral, and DeepSeek's open releases are prominent examples. This contrasts with closed models like GPT and Claude, which are accessed only through an API.
Open weights bring real advantages: full control, privacy (data never leaves your environment), no per-token API fees, and the ability to customize deeply. The costs are that you provide the infrastructure and expertise to run them, and the very top of the capability frontier is often a closed model.
Note the term: 'open-weights' means the parameters are public, which isn't the same as fully open-source (training data and code are often not released). It's a spectrum, not a binary.