Despite heavy government regulations and ongoing struggles with hardware sourcing, Tencent’s large language model has been released for business use in China.
Chinese technology corporation Tencent unveiled its “Hunyuan” artificial intelligence (AI) system, a multimodal large language model (LLM) similar to OpenAI’s ChatGPT, at the Global Digital Ecosystem Summit in Shenzhen on Aug. 7.
Say hi to Hunyuan! Our large foundation model, unveiled at the Global Digital Ecosystem Summit. The platform is now available to enterprises in China, supporting a heap of functions, from image creation, copywriting, customer service and more. https://t.co/ifLfc9yCVQ #AI pic.twitter.com/wWhRfzl3km
— Tencent 腾讯 (@TencentGlobal) September 7, 2023
Hunyuan marks Tencent’s entry into “foundational models,” so named for their development as a framework upon which associated AI APIs can run.
According to a blog post from Tencent, Hunyuan is intended to function as a comprehensive suite of AI tools:
“Tencent’s foundation model supports a wide array of functions spanning the creation of images, copywriting, text recognition, and customer service, to name a few. These will be instrumental in key industries like finance, public services, social media, e-commerce, transportation, games, and many more.”
The AI system has also been integrated into Tencent’s ecosystem of applications and services with connectivity to Tencent Cloud, Tencent Marketing Solutions, Tencent Games, Tencent fintech services, Tencent Meeting, Tencent Docs, Weixin Search, and QQ Browser.
The launch comes as relations between the U.S. and China remain icy after the Biden administration signed off on an export ban on certain kinds of computer chips, including hardware commonly used to develop and train AI systems, in October 2022.
According to a recent report, the two administrations continue to engage in diplomatic talks but haven’t made any headway towards establishing a timeline for ending the ban.
Tencent, for its part, says the launch of Hunyuan signals that it’s “committed to open collaboration in the ecosystem, with domestic businesses ultimately benefiting from the company’s high-quality model services, while international businesses leverage Tencent to access the Chinese market.”
Currently, ChatGPT isn’t available in China. Though Hunyuan isn’t the first LLM launched in the Chinese market, as Cointelegraph reported, it follows scores of other models from the likes of Alibaba and Baidu, it might be the most significant.
Tencent Holdings is China’s biggest technology company. Its products have nation-wide reach and, due to the ubiquity of WeChat and its associated apps, Hunyuan has the potential to permeate the domestic market in ways that OpenAI and Microsoft might not be able to imitate with ChatGPT in the west.
Hunyuan is supposedly similar in power and capability to GPT3 (OpenAI’s core model circa 2022) when it comes to raw numbers as reported by Tencent. At 100-billion parameters and two-trillion tokens, the LLM takes its place as one of the most powerful LLMs in the world.
However, it’s worth noting that parameter and token counts aren’t necessarily indicative of the capabilities of a multimodal AI system. Hunyuan has the benefit of being trained on a massive corpus of Chinese language text. Theoretically speaking, this would give it an advantage over models trained primarily on non-Chinese texts when it comes to operations in the Chinese language environment.