TokenAI launches compact AI decision model

TokenAI, an Alexandria-based startup, has launched Neo, an open-weights decision model designed to route tasks to the appropriate tools or escalate them for human review. The model features 41 million parameters and was trained from scratch on 400,000 synthetic records covering a million independent decisions. Neo produces three types of outputs in a single pass: a choice among up to 32 candidate options, a score across 5 ordered levels, and a binary yes/no decision. The model achieved an accuracy of 73.99% on its scoring head and 82.38% on its binary decision head, making it an effective tool for enhancing routing logic within AI systems.