Ant Digital launches LingDT-Decision-1.0 decision models
Ant Digital Technology released LingDT-Decision-1.0 in 4B and 27B versions for structured enterprise tasks such as intent recognition, classification, tool selection, routing, and risk assessment. The models add native multi-select and dependency-aware DAG sequencing, and are available through the DTMaaS platform with a limited free trial. Ant Digital reported lower token use, faster processing, and leading scores across 23 open-source benchmarks, but the supplied reports contain no independent validation of those performance claims.
InfoQ notes that decision models such as LingDT-Decision-1.0 offer enterprises another option for high-frequency, structured judgments that may otherwise be handled by general-purpose language models.
10月22日，蚂蚁数科发布决策模型 LingDT-Decision-1.0，提供4B和27B两个参数规模版本，主要面向智能客服、安全风控、金融等企业应用场景。与侧重文本生成的通用大语言模型不同，该模型主要处理意图识别、任务分类、工具选择及风险判断等结构化决策任务，目前已上线蚂蚁数科 DTMaaS 平台，开发者可限时免费体验。
根据蚂蚁数科公布的测试结果，在部分企业业务场景中，LingDT-Decision-1.0相比所选通用大语言模型，Token 消耗降低超过90%，任务处理速度最高提升约10倍。在覆盖23项开源基准的评测中，其27B版本的综合评分超过了Microsoft-Decision-1、Jev 1.13.0及Fastino/GLiDE等决策模型。
#Ant Digital Technology#LingDT-Decision-1.0#DTMaaS#Microsoft-Decision-1#Jev 1.13.0#Fastino/GLiDE#DeepSeek-V4-flash#Qwen-3.7-flash