Tencent Holdings announced on Thursday (7 Sept) that its large language model (LLM) “Hunyuan” has debuted for enterprise use.
Tencent demonstrated the AI model to a live audience at a Shenzhen conference showing how it is capable of conversing in both Chinese and English.
Go deeper with GlobalData
With Hunyuan’s launch, Tencent joins a number of other Chinese companies to release AI models in recent months. Baidu, China’s largest search engine, released its AI chatbot, Ernie 3.5, following regulatory approval on 5 September.
Baidu CEO, Robin Li, disclosed at an industry event last week that over 70 LLMs are now available in China.
Baidu has claimed its ChatGPT rival, Ernie, had outperformed OpenAI’s ChatGPT chatbot in a series of tests.
Hunyuan has over 100 billion parameters and was trained with 2 trillion tokens, Reuters reported. For comparison, OpenAI’s GPT-3 AI model had 175 billion parameters in 2020 and Meta’s Llama 2 model had 70 billion parameters in 2023.
Tencent claimed that Hunyuan has 30% fewer hallucinations than Meta’s Llama-2, though these claims have not been independently verified. AI hallucination is the scenario in which an LLM makes up false information which is not based on real data or events.
Earlier this year, Google’s Bard, notoriously made false claims during its press demo.
