On April 14th, at the 2025 SenseTime Technology Exchange Day held in Beijing, SenseTime officially announced a comprehensive upgrade to its SenseCore 2.0 large-scale AI infrastructure. As a leader in AI infrastructure, SenseCore 2.0 aims to provide businesses with agile, flexible, and reliable full-stack AI infrastructure services, driving the efficient deployment and large-scale application of large models with exceptional cost-effectiveness.
The SenseCore 2.0 upgrade addresses three major challenges facing the current large model industry: the mainstreaming of open-source models and tools, the critical issue of cost-effectiveness, and the new scenario needs brought about by multi-modal technologies. Through technological innovation, SenseCore achieves a significant improvement in the utilization rate of heterogeneous computing power, dramatically reducing inference costs. Online inference performance is improved by 15% compared to leading vendors, while offline inference is 5 times faster in the Prefill stage and 3.5 times faster in the Decode stage. These breakthroughs provide ultra-cost-effective infrastructure services for large model applications.
SenseCore actively embraces the open-source ecosystem. Through OpenAPI compatibility, K8S native platform adaptation, open-source large model hosting services, a full-stack open-source toolchain, and a wealth of open-source components, it achieves seamless integration and agile iteration of the technology stack. SenseCore provides a one-stop model service with "zero migration cost, zero threshold for use, and zero development and deployment," catering to businesses' needs from standardized applications to personalized customization.
Furthermore, SenseCore has launched four industry solutions targeting embodied intelligence, AIGC, AI4S (AI For Science), and the intelligent transformation of traditional enterprises. These solutions cover the entire chain from scenario definition to business implementation, providing end-to-end support to customers and solving the "last mile" problem of intelligent implementation. SenseTime and Songying Technology have signed a strategic cooperation agreement to collaborate comprehensively on data simulation, R&D pipelines, and customer service to jointly promote the development of embodied intelligence technology.
To further accelerate the full-scenario implementation of AI, SenseTime announced an investment of 100 million yuan in special vouchers, focusing on embodied intelligence, AIGC, and multiple industries. This provides businesses with full-process AI service support, from expert consultation to model training and inference verification. This initiative aims to create a "high-efficiency, low-cost, end-to-end" one-stop empowerment solution to help customers in various industries accelerate the commercialization of large models.
