SenseTime Among the Top Three for China’s Large Model Platform Market Share

International Data Corporation (IDC) recently released the "China Large Model Platform Market Share, 2024" report.

According to the 2024 data, SenseTime ranks third in China's large model development platform market with a 12.2% market share, firmly maintaining its position in the top tier of the industry.

SenseCore Wanxiang, Driving Intelligent Enhancements

IDC said that over the past year, the focus of competition among large model platforms has increasingly shifted towards intelligent agent development platforms. Companies are now competing on the comprehensiveness of their tool chains, the use of tokens, and the acquisition of key customers.

The report defines a large model platform as a full-lifecycle management tool that supports training, tuning, inference, and model deployment. As a leading representative in this field, SenseTime has developed the Wanxiang Large Model Development and Service Platform, integrating model deployment, inference optimization, and application development, deeply integrating SenseNova with open-source ecosystem capabilities. Its core strengths include:

Multimodal Interaction: Supports input and output across three modalities – voice, video, and text.

Efficient Inference Engine: Equipped with a multimodal hybrid inference engine to optimize processing efficiency.

Model Lightweighting: Provides model compression technology to reduce deployment costs and barriers.

Enterprise-Grade RAG: Ensures accuracy, security, and controllability of knowledge retrieval.

Furthermore, SenseCore Wanxiang powers AI solutions for various industries, including embodied intelligence, AIGC, AI4S, and intelligent transformation in traditional sectors, enabling end-to-end deployment from scenario definition to business implementation.

Industry Implementation: Wanxiang Platform Accelerates Value Realization

Currently, the SenseCore Wanxiang Large Model Development Platform has been implemented across various industries and business scenarios, including surveying and design, urban governance, smart ports, smart homes, and healthcare.

In the surveying and design field, SenseCore collaborated with China Railway First Survey and Design Institute Group (FSDI) to develop a multimodal large model application platform for national railway value engineering. This initiative marks the first application of multimodal large models in the intelligent utilization of railway engineering design knowledge. By integrating heterogeneous data such as text, drawings, and formulas, the platform offers five core functions – knowledge retrieval, intelligent Q&A, text generation, text review, and multimodal dialogue – significantly enhancing the efficiency and intelligence of railway engineering design.

In urban governance, Sensetime has played a key role in supporting the Shanghai Municipal Planning and Natural Resources Bureau's Cloud-Cosmos Industry Foundation Large Model. The company provided an integrated solution encompassing the Wanxiang large model service platform and the SenseNova multimodal large model. This model integrates data from territorial spatial planning and natural resources, including academic materials, policies and regulations, technical standards, and approval records. It can integrate and interpret multimodal data such as satellite imagery and planning drawings, delivering intelligent support for urban spatial governance and scientific decision-making.

In public services, SenseCore has empowered a government agency to build a compact AI innovation platform. Leveraging the Wanxiang model development platform along with supporting services such as model optimization, API calls, deployment, and monitoring, the solution significantly streamlines complex deployment processes. It offers an "out-of-the-box" experience, significantly lowering the barrier to implementing AI applications in government scenarios.

In its report, IDC emphasized that the future of large model platforms will not only revolve around model capabilities but will also be about toolchains, ecosystems, and industry implementation.

Looking ahead, Sensetime will continue to leverage the robust AI infrastructure of SenseCore, the ongoing iteration of the SenseNova large model, and the innovative capabilities of the Wanxiang platform to serve more industry customers. By delivering more efficient AI capabilities and more accessible computing resources, SenseTime aims to provide full-stack empowerment for the large-scale implementation of AI technology.