Recently, the MaaS Model as a Service Technology and Application Requirements Part 4: Model Service Platform, co-authored by iSoftStone and led by the MaaS working group of the China Artificial Intelligence Industry Development Alliance (AIIA), was officially released. The AIIA, initiated by the CAICT, includes contributions from various research institutions, universities, enterprises, and social organizations, such as the China Electronics Standardization Institute, the National Industrial Information Security Development Research Center, the Institute of Automation of the Chinese Academy of Sciences, the Fourth Electronics Research Institute, the First Electronics Research Institute.

Starting in 2023, the AIIA formed the MaaS working group to collaborate with industry stakeholders on the development of a series of MaaS standards, addressing challenges in the implementation of MaaS. The research focuses on five dimensions: model customization, model service deployment, model management, model application development, and model service capabilities, providing references for capability building for large model service providers and platform suppliers, as well as technical selection guidance for application parties.

The MaaS Model as a Service Technology and Application Requirements Part 4: Model Service Platform primarily targets both model service platform builders and users, outlining four key capabilities required for deploying models as services: model operation acceleration, model service deployment, model service testing, and model service management operation. It also standardizes the full-process capabilities for transforming model services into APIs, guiding model service platform suppliers to enhance their capability construction, optimize deployment efficiency, and improve service quality.

As a leader in digital technology products and service innovation in China, iSoftStone has prioritized artificial intelligence (AI) in its strategic development, gradually building full-stack AI service capabilities. The company integrates hardware products such as AI servers, AI workstations, and AI PCs from iSoftStone Computing (Tongfang Computer) and AIStone Global (Tongfang International) to create a collaborative ecosystem for AI hardware and software. At the infrastructure level, iSoftStone continuously enhances AI large model training, inference, and optimization capabilities based on its intelligent cloud platform. At the platform level, the company focuses on developing new intelligent technology platforms, including the AISE software engineering platform, RAG Plus AI knowledge middle office, and Tianfang industrial internet AI platform. At the application level, iSoftStone leverages consulting services to support innovative AI and large model application scenarios, facilitating the rapid implementation of large model applications for enterprises. To date, iSoftStone has successfully implemented large models in sectors such as insurance, retail, healthcare marketing, and industrial knowledge management based on iSoftStone Tianxuan MaaS platform.

Since 2024, iSoftStone has actively participated in the formulation of industry standards for artificial intelligence data infrastructure, large model infrastructure, MaaS, and large models in finance and automotive sectors. This involvement not only enhances the top-level design capabilities of iSoftStone's AI products and services but also contributes to the development of its ecosystem. Furthermore, the company aims to drive large-scale application and commercialization of AI technologies in the industry through standardization. Looking ahead, iSoftStone will continue to deepen collaboration with industry partners, promoting the standardization and industrialization of AI technologies, and providing more mature, stable, and reliable AI solutions for the industry.