NEX AGI's Nex N2.5 Max is the text-only member of the family, built on a 1.6-trillion-parameter 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 → foundation. The vendor describes it as its first complete post-training effort at trillion-parameter scale. This means the family's discussion of visual computer interaction should not be assumed to apply to Max. Its model card instead documents a deployment example spanning two nodes and 16 H200 GPUs. Max's release is therefore relevant to teams evaluating large-scale text models and the infrastructure needed to host them.
Terms in this piece · Glossary
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.
