Ant Digital Technologies has released LingDT-Decision-1.0, a model designed for structured business decisions that the company says cuts token consumption by over 90% compared to large language models.
The model comes in 4B and 27B parameter versions and is available on Ant Digital's DTMaaS platform, where developers can try it free for a limited time.
Ant Digital says the model is the first to support native multi-select and sequential decisions arranged as a directed acyclic graph (DAG), which fits multi-step business processes.
In tests across customer service, risk control and financial scenarios, the company says LingDT-Decision-1.0 cut token consumption by more than 90% compared with large language models, responded up to 10 times faster and performed better.