China’s Tencent Releases Large Language Model, Opens it For Enterprise Use

TENCENT’S LARGE LANGUAGE MODEL ‘HUNYUAN’, which can converse in both Chinese and English, is supposedly superior to OPENAI’s CHATGPT in a number of respects.
After receiving public release permits for AI from Chinese authorities last month, Tencent has now entered the competitive global AI race with the release of its large language AI model Hunyuan.

The WeChat parent company unveiled the ‘Hunyuan’ AI model in front of a packed conference room in Shenzhen, with the stated goal of becoming China’s AI leader. Officials from Tencent claimed that its AI model will form the backbone of more than 50 of the company’s products and services during the launch.

Jiang Jie, vice president of Tencent, said, “By July, there are more than 130 large language models in China.” “A war of a hundred models has begun,” he continued.

Now available for business usage, Tencent’s Hunyuan follows the release of similar huge language models and AI solutions by other Chinese tech giants including Baidu Inc and SenseTime Group.

Tencent reportedly used over 2 trillion tokens during Hunyuan’s training, giving the massive language model around 100 billion parameters.

Meta’s most recent artificial intelligence model, Llama 2, had more than 70 billion parameters, whereas OpenAI’s previous GPT-3 model has approximately 175 billion.

With the ability to converse in both Chinese and English, Tencent’s huge language model ‘Hunyuan’ is said to be superior to OpenAI’s ChatGPT in a number of ways.

Besides its comparison with ChatGPT, Tencent has claimed a number of other hallucination-related claims, none of which have been objectively validated.