Model families
What is a model family, and what does open weight mean?
A model family is a line of related models from a single maker, sharing a name and design lineage. The main split in 2026 is between frontier models you access over an API and open-weight models whose files you can download and run yourself.
A model family is a line of related AI models that come from one maker and share a name, a design lineage, and usually a release rhythm. Claude is a family from Anthropic. GPT is a family from OpenAI. Within a family you find versions over time (a newer release replaces an older one) and sizes at a moment (a small, fast model next to a large, capable one). When people say “which model should I use,” they almost always start by picking a family, then a size and version inside it.
The clearest way to sort the 2026 landscape is by how you get access. Frontier models are the largest, most capable releases, and you reach them over the internet through the maker’s service. You rent the use; you never hold the model file. Open-weight models are different: the maker publishes the trained numbers (the weights) for download, so you can run the model on your own hardware, inspect it, and adapt it.
Open weight is not the same as fully open source. A maker can release the weights while keeping the training data and recipe private, and the license may still restrict some uses. The practical point for a beginner is the divide itself: some families are services you call, and some are files you can keep. The two pages that follow walk through each side.