Language Model
A model trained on text, whose basic skill is predicting the most likely next word (or character) given some input. That one ability, at large scale, is what underlies translation, chatbots, autocomplete, and most of the tools discussed on this wiki.
Because it predicts text rather than looking answers up, a language model is probabilistic: the same prompt can produce different answers, and it can state a wrong answer as fluently as a right one (see Ojibwe Chat). It has no built-in notion of being correct, only of being plausible.
Language models are often described by size. A small one can run on ordinary, local hardware. The largest need data centers and are typically reached through a company's closed-weight service. That difference is not just technical: it shapes cost, who controls the data sent to the model, and whether a community can run the tool on its own terms.