Meet the 'Six Tigers' that dominate China's AI industry

Zhipu AI was founded in 2019 out of Tsinghua University by two professors, and is one of China’s earliest generative-AI startups. The Beijing-based company develops foundation models that power its applications, including a conversational chatbot called ChatGLM, and an AI video generator, Ying.

In August, the startup introduced its GLM-4-Plus model, which it said performs on par with OpenAI’s GPT-4o. GLM-4-Plus was trained on high-quality synthetic data, and can process large amounts of text. Zhipu released its GLM-4-Voice end-to-end speech model in October, which has human-like speech capabilities such as intonation and dialect. The model can engage in real-time voice conversations in Chinese and English.

In January, the outgoing Biden administration added Zhipu to its restricted trade list along with more than 20 other Chinese firms suspected of aiding China’s military.

Zhipu raised more than one billion yuan — about $140 million — in a financing round earlier this month that included Alibaba (BABA), Tencent (TCEHY), and some state-backed entities.

Moonshot AI was also founded in 2023 at Tsinghua University by Yang Zhilin, an AI researcher and alumnus of both Tsinghua and Carnegie Mellon University.

The Beijing-based startup’s Kimi AI chatbot is one of China’s top five AI chatbots, and had almost 13 million monthly active users as of November, according to Counterpoint Research. Kimi can process queries of up to two million Chinese characters.

The company, which is valued at $3.3 billion, is backed by some of China’s largest tech firms, including Alibaba and Tencent.

MiniMax was founded in 2021 by AI researcher and developer Yan Junjie, and developed the popular AI chatbot, Talkie. The app, which was launched as Glow in 2022, was improved and rebranded to Xingye in China, and Talkie in international markets where it’s available.

Talkie allows users to chat with various characters, including celebrities and fictional characters. According to the South China Morning Post, Talkie was removed from the U.S. Apple App Store in December, citing “technical reasons.” The Shanghai-based company also developed a text-to-video AI generator called Hailuo AI.

Last March, Alibaba led MiniMax’s $600 million funding round, which led to a $2.5 billion valuation.

Baichuan Intelligence was founded in March 2023, and counts talent from Microsoft (MSFT) and Chinese tech giants such as Huawei, Baidu (BIDU), and Tencent.

The Beijing-based company developed two open-source large language models, Baichuan-7B and Baichuan-13B, which it released in 2023. The AI models are commercially available in China and were tested on Chinese, English, and multi-language datasets for general knowledge, mathematics, coding, language translation, law, and medicine.

In July, Baichuan raised five billion yuan, or about $687.6 million, in a funding round valuing the company at more than 20 billion yuan. Alibaba, Tencent, and some state-backed funds were among the investors.

StepFun has released 11 foundation models, including visual, audio, and multimodal AI systems. The Shanghai-based company was founded in 2023 by Jiang Daxin, a former senior vice president at Microsoft.

The startup’s Step-2 language model has one trillion parameters and is ranked among competing models from companies such as DeepSeek, Alibaba, and OpenAI on LiveBench, which benchmarks large language models.

In December, Fortera Capital, a state-owned private equity firm, helped Stepfun raise “hundreds of millions of dollars” in Series B funding.

01.AI was founded by Kai-Fu Lee, a veteran of Apple (AAPL), Microsoft, and Google, in 2023. The Beijing-based company has launched two models, Yi-Lightning and Yi-Large. Both AI models are open-source, and are among the top-ranked large language models in the world for language, reasoning, and comprehension.

The Yi-Lightning model stands out for its efficient training costs. On LinkedIn, Lee said that Yi-Lightning was trained on 2,000 of Nvidia’s H100 chips for one month — far fewer chips than xAI’s Grok 2, which it performed comparably with. Yi-Large, meanwhile, can engage in human-like conversations in both English and Chinese.