Arcee AI's December 2025 Trinity announcement separated two compact models by intended use. Mini was a reasoning modelA model trained to think — generating extended internal reasoning before answering — trading time and tokens for accuracy on hard problems.Full definition → for tools and multistep tasks, with 26 billion parameters but only about 3 billion active for each generated 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 →. Its mixture-of-expertsA model built from many specialist sub-networks where only a few activate per token, giving big-model capability at small-model running cost.Full definition → design selects part of the model for each step, separating total model capacity from the computation used at that step. Nano Preview targeted lightweight conversation and instruction following and was explicitly described as experimental. Those roles matter when interpreting the family's promise of efficient deployment. Trinity manifesto
Terms in this piece · Glossary
reasoning model — A model trained to think — generating extended internal reasoning before answering — trading time and tokens for accuracy on hard problems.
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.
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.
