Recently, Zhipu AI released customized personal GLM large model GLMS and GLMSTORE, which are fully benchmarked against OpenAI's GPTS, and users can create their own large model applications through simple prompt word commands. This is an important breakthrough in the field of large models in China, which is expected to accelerate the commercialization process of domestic AI and promote the recovery of the sector. The Artificial Intelligence ETF (**159819, Connection**a C:012733 012734) is a one-click configuration of core leading companies in subdivided industries such as computing chips, AI large models, and AI applications.
Zhipu AI has released a new generation of pedestal model GLM-4 - the performance is fully comparable to GPT-4. Among them, GLM-4 not only greatly improves Chinese ability. Moreover, the ability to write long texts is also unbeatable. Specifically, GLM-4 can support a context window length of 128K; A single prompt can process as many as 300 pages of text. At the same time, in terms of multimodal capabilities, both Wensheng diagram and multimodal understanding have been enhanced. In addition, Zhipu AI also released customized personal GLM large models GLMS and GLMSTORE this time, which are fully benchmarked against OpenAI's GPTS.
In addition to Zhipu, the overall level of domestic large models is also continuing to upgrade. According to Superclue evaluation, in terms of Chinese ability, the average level of representative large models in China has continued to improve since last year, and has now exceeded GPT-35, and constantly approaching GPT-40。Looking forward to 2024, driven by the open source ecology of overseas large models and abundant data resources in China, the upgrade of domestic large models is expected to accelerate.
With the accelerated upgrading of large models, domestic high-quality AI applications are expected to land, driving the recovery of the sector. The commercialization process of overseas AI is leading, and domestic high-quality AI applications have not yet been formed, which has caused a certain gap between domestic and foreign stock prices, but at the same time, it reflects that there is still a large space for domestic AI. The birth of high-quality AI applications in China requires sufficient computing power support and an excellent large model base. In terms of large models, the performance of domestic large models has been shrinking with the international leading level; In terms of computing power, the performance of domestic computing power chips has made breakthroughs, and the AI chips accumulated in the early stage are also relatively sufficient, which has laid a solid foundation for the launch of domestic high-quality AI applications. 2024 is expected to be the first year of domestic AI applications, and the beta of the AI sector is worth paying attention to.
The Artificial Intelligence ETF (**159819, Connect**a C:012733 012734) closely tracks the CSI Artificial Intelligence Thematic Index.
The above content and data have nothing to do with the position of the interface and do not constitute investment advice. Do so at your own risk.