Currently, the focus of the competition among large models has shifted from simply comparing model performance to how to achieve large-scale application in real enterprise scenarios with lower costs and higher efficiency. Understanding enterprises better and meeting their needs more effectively has become a key focus of the industry.
On April 29, Ant Digital announced the launch of the commercial version of Ling-2.6-flash, LingDT-2.6-flash. This model emphasizes "Token efficiency," saving 90% of Token consumption compared to models of the same parameter level under the same evaluation tasks, making it faster, more cost-effective, and more suitable for large-scale deployment in real enterprise scenarios.
According to the information, LingDT-2.6-flash fully inherits the core advantages of Ling-2.6-flash, optimizing Token efficiency through low-level architecture optimization, solving the challenges of industrial implementation.
This model has a total parameter count of 104B and an activated parameter count of 7.4B. It significantly improves inference efficiency and Token utilization while maintaining a high level of intelligence. Additionally, LingDT-2.6-flash is equipped with a financial-grade security protection system, supporting private deployment, data isolation, and API permission control, fully meeting enterprises' strict data security and compliance requirements, achieving a balance between high performance and high security.
Currently, the model has been launched on Ant Digital's DT MaaS platform, offering a limited-time 80% discount. The cost for inputting one million tokens is only 0.48 yuan, and outputting one million tokens is 1.44 yuan, greatly reducing the cost of large-scale AI application for enterprises.
Ant Digital has been deeply focusing on industrial scenarios, adhering to the AI technology roadmap of "high performance, low cost, and strong practicality." It has built a full-stack AI product matrix covering the underlying large model, model service layer, intelligent agent platform, and upper-layer intelligent agent applications. Relying on its mature engineering capabilities, it has brought AI from technical demonstrations to the front lines of business, accelerating its implementation in key areas such as finance, energy, and transportation.
For example, in the financial sector, Ant Digital has successively launched the financial reasoning large model Agentar-Fin-R1 and the lightweight MoE model LingDT-Fin-Mini 2.5. Through the collaborative model of "industry brain" plus "executive limbs," it provides AI large model solutions for the financial industry that are professional, cost-effective, and capable of rapid response. Currently, Ant Digital has launched over 100 intelligent agent applications in the financial field, including AI mobile banking, insurance claims, security lockdown, and wealth management, among other core application scenarios.
