Ant Digital Releases LingDT-Decision-1.0 Model, Cutting Token Use by 90%

Ant Digital has released LingDT-Decision-1.0, a structured decision-making model available in 4B and 27B versions designed for rapid industrial applications. The model reduces token consumption by over 90% and improves response speeds by 10 times compared to traditional large language models. Traditional LLMs are inefficient for high-frequency business tasks because they spend compute generating and parsing text sequentially. By focusing directly on structured judgments, this model drastically lowers operating costs and latency for enterprise applications like customer service and risk control. LingDT-Decision-1.0 is the first decision model to natively support multi-select and sequential decision-making (DAG) capabilities. In 23 open-source evaluation benchmarks, the 27B model achieved top accuracy among models of a similar size, outperforming popular models like Microsoft-Decision-1 and Jev 1.13.