The launch of ChatGPT-3.5 in December 2022 triggered a global surge in interest in large-scale AI models, with major Chinese companies such as Baidu, Alibaba, and SenseTime following suit in 2023. By June, a flurry of general-purpose and specialized models had emerged, contributing to China’s rapidly growing AI landscape. By the end of 2023, over 300 large models had been released, leading to what the media dubbed the “Hundred-Model Battle.” As of 2024, the so-called “Six Tigers”-Zhipu AI, MiniMax, Baichuan AI, Moonshot, StepFun, and 01.AI-are dominating the field, with a valuation at over $1 billion each, and four of them exceeding 20 billion RMB ($2.7 billion), according to Chinese media outlet Caixin. These companies are also bearing the weight of expectations to become China’s version of OpenAI.
Zhipu AI
Founded in June 2019, Zhipu AI is one of the few companies in China that predates the launch of ChatGPT. As one of the first companies to explore large language models, Zhipu AI has become a major player in the sector. The company’s leadership team boasts notable figures from China’s top academic institutions. CEO Zhang Peng, a graduate of Tsinghua University’s Computer Science Department, is joined by Chief Scientist Tang Jie, a professor at Tsinghua University, and President Wang Shaolan, a Tsinghua Innovation Leadership Doctorate holder. The company also expanded its leadership team in April 2024, with former Alibaba Cloud Vice President Chen Xuesong joining as Vice President.
In October 2024, the company released GLM-4.0, an open-source end-to-end speech large language model capable of replicating human-like interactions and offering users the ability to adjust the tone, emotion, or dialect according to their preferences. Zhipu has also launched a wide range of other products related to large models, including various general-purpose models of different sizes, the CodeGeeX code model based on Huawei’s MindSpore framework, the CogView text-to-image model, and the VisualGLM multimodal dialogue model. In an announcement on December 31, 2024, the GLM technology team revealed plans to enter the field of inference models.
On the financial front, Zhipu AI has been a favorite among investors. In 2023, the company completed an RMB 2.5 billion ($350 million) funding round, with participation from major investors including the Social Security Fund’s Zhongguancun Innovation Fund, Meituan, Ant Group, Alibaba, Tencent, Xiaomi, Sequoia Capital, and Hillhouse Capital. In another boost to its financial standing, the company secured a new round of RMB 3 billion ($420 million) in funding by the end of 2024.
MiniMax
MiniMax was founded in 2021- two years after its predecessor, ZhiPu. Before founding MiniMax, Yan Junjie, the company’s founder, spent over a decade in AI research and development. At MiniMax’s first offline event in August this year, Yan shared the company’s founding vision-Intelligence with Everyone, and revealed key user data: 3 billion daily AI interactions and 30 trillion tokens processed. In October 2022, MiniMax launched its first app, Glow, which quickly gained traction, surpassing 5 million users within just four months. The app was later rebranded as Xingye for the Chinese market and Talkie for international users, marking the company’s expansion into global markets.
Another standout product is Hailuo AI, a video-generator that quickly gained international attention later, with video creators offering positive reviews. In March 2024, Alibaba led a $600 million funding round for Minimax.
Baichuan Intelligence
Baichuan Intelligence was established in March 2023, with its core team consisting of top AI talents from renowned tech companies such as Sogou, Baidu, Huawei, Microsoft, ByteDance, and Tencent. Less than 100 days after its founding, Baichuan Intelligence released two open-source, commercially usable Chinese language models, Baichuan-7B and Baichuan-13B.
In July 2024, Chinese media outlet 36kr reported that Baichuan Intelligence completed a Series A funding round, raising a total of 5 billion yuan. Previously, in October 2023, Baichuan had announced its A1 funding round, revealing a list of investors that included tech giants Alibaba, Tencent, Xiaomi, and several top-tier investment firms.
Moonshot
Moonshot, also founded in March 2023, is led by Professor Yang Zhilin from Tsinghua University’s School of Interdisciplinary Information. The team includes talent from global tech giants such as Google, Meta, and Amazon. Unlike some of the large models currently on the market that aim for a one-size-fits-all approach, Moonshot’s large model, Kimi Chat, is more focused on long-text capabilities. For instance, its practical use can support a context of approximately 200,000 Chinese characters—2.5 times the size of Anthropic’s Claude-100k (which supports around 80,000 words) and 8 times that of OpenAI’s GPT-4-32k (which supports around 25,000 words). This potential has contributed to Moonshot’s valuation of $3 billion.
However, its success is recently overshadowed by legal disputes.
StepFun
StepFun is the latest company to officially debut among the “Six Little Tigers.” Many people only heard of Jieyue for the first time in March this year, but the company quickly rose to become a leading player in the large model sector. Its strong multimodal technology has garnered widespread attention from developers. According to statistics, in the past 10 months, Jieyue Xingchen has launched a total of 11 foundational model products, including language models, multimodal understanding models, video generation models, and speech models.
01.AI
01.AI was founded in July 2023 by Kai-Fu Lee, who held key positions at tech giants such as Apple, Silicon Graphics, Microsoft, and Google. In the first three quarters of this year, 01.AI launched two models—Yi-Lightning and Yi-Large—both of which ranked among the top on LMSYS and were the highest-ranking domestic large models. However, the core advantage of 01.AI lies not in the rankings, but in the cost of pre-training. Under the model co-building strategy, the training and deployment costs are impressively low. Founder Kai-Fu Lee stated that Yi-Lightning’s training only required 2,000 GPUs and took just one and a half months, with costs amounting to around 2% of those of Musk’s xAI.
On August 7, 2024, media reports revealed that 01.AI had completed a new funding round, raising several hundred million dollars. Investors in the round included an international strategic investor and several Southeast Asian consortiums. 01.AI has yet to comment on the funding news. Before this round, 01.AI had previously received investment from Alibaba Cloud, following which, the company’s valuation surpassed $1 billion, securing its position as a unicorn.
