Tencent, a major player in China’s artificial intelligence (AI) market, has joined the trend of offering its foundation model for organizations to test and develop their own applications. The model, called Hunyuan, can be accessed through APIs on Tencent’s cloud platform and can be customized to meet specific requirements. It offers capabilities such as Chinese language processing, advanced logic reasoning, image creation, text recognition, and copywriting. Tencent is promoting Hunyuan as a platform for various industries, including finance, e-commerce, transport, and games. The AI model currently has over 100 billion parameters and more than two trillion tokens in pre-training data. Tencent has integrated Hunyuan with 50 of its own products, including fintech, advertising, games, office productivity, and search applications.
One example of Hunyuan’s application is in Tencent Meeting, where an AI assistant can generate meeting minutes using natural language processing and user prompts. The advertising tools can also be used to create shopping guides for retailers to use as marketing assets. Dowson Tong, Tencent’s senior executive vice president of cloud and smart industries, stated that the launch of Hunyuan aims to balance the performance of consumer-facing AI chatbots with the need for businesses to increase efficiency, reduce costs, and protect privacy and proprietary data.
Tencent is part of a growing list of Chinese companies that are contributing to the generative AI market, which includes Baidu, JD.com, and Alibaba Cloud. JD.com’s ChatRhino, for example, has a base of 100 billion parameters, surpassing its previous model Vega, which led the General Language Understanding Evaluation (GLUE) list. Alibaba Cloud offers Tongyi Qianwen for beta testing to domestic customers and developers through an API. The company also introduced a partnership program to foster the development of AI applications in various industries, including finance and petrochemicals.
China’s push towards AI development is accompanied by interim regulations aimed at ensuring the healthy growth of the technology and safeguarding national security and public interests. Effective from August 15, these regulations outline measures to improve the quality of training data, including accuracy, objectivity, and diversity. Generative AI service providers are also held legally responsible for the information generated and its security, and they must establish service-level agreements with users to clarify rights and obligations.
Overall, Tencent’s introduction of Hunyuan as a customizable AI model reflects the company’s commitment to providing solutions that cater to the specific needs of Chinese enterprises across various industries. With the growing competition in China’s generative AI market, it will be interesting to see how Tencent and other players continue to innovate and differentiate their offerings.