mixture-of-experts — A model built from many specialist sub-networks where only a few activate per token, giving big-model capability at small-model running cost.
pretraining — The first, biggest phase of building a model: training it on enormous amounts of text so it learns language, facts, and reasoning in general.
open weights — A model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
Why it matters
Trinity Nano and Mini ship as open weightsA model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.Full definition → with an American end-to-end training provenance, and Nano pushes sparsity to 800M active parameters per tokenThe chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.Full definition → across 128 experts. Mini is the post-trained reasoning variant; Nano Preview is experimental and not API hosted.
