Yan Junjie
Last reviewed
Jun 8, 2026
Sources
13 citations
Review status
Source-backed
Revision
v1 · 1,628 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 8, 2026
Sources
13 citations
Review status
Source-backed
Revision
v1 · 1,628 words
Add missing citations, update stale details, or suggest a clearer explanation.
Yan Junjie (Chinese: 闫俊杰; born 1989) is a Chinese computer scientist and technology entrepreneur who is the founder, chairman, chief executive officer, and chief technology officer of MiniMax, a Shanghai-based artificial intelligence company counted among China's so-called "AI tigers." Trained as a computer vision researcher, he spent roughly six years as a senior research executive at SenseTime before leaving in December 2021 to build MiniMax around the goal of artificial general intelligence. Under his leadership MiniMax developed the abab and MiniMax-M families of large language models, the Hailuo video and audio generators, and the Talkie and Xingye companion apps, and in January 2026 it listed on the Hong Kong Stock Exchange in one of the fastest founding-to-listing runs in the AI industry, a debut that made Yan a billionaire at the age of 36. [1][2][4]
Yan was born in 1989 in Xiayi County, Shangqiu, in China's Henan province. [1][3] He graduated from Southeast University in Nanjing in 2010 with a bachelor's degree in mathematics. [3] He then pursued graduate study at the Institute of Automation of the Chinese Academy of Sciences, where he worked in the National Laboratory of Pattern Recognition and completed a PhD in artificial intelligence, focused on computer vision, in July 2015 at the age of 26. [1][3] He subsequently carried out postdoctoral research in deep learning and computer vision in the computer science department of Tsinghua University. Over his research career he has been credited with roughly 200 papers at top conferences and journals and more than 30,000 citations. [3]
By his own later account, Yan's ambitions as a doctoral student were modest: he has said he once aspired simply to become a Java developer at IBM earning an annual salary of about 280,000 yuan. [1] An avid Dota 2 player who used the in-game handle "IO," a pun on the input and output of a computer program, he has traced his pivot from computer vision toward language and AGI to 2019, when he watched OpenAI's bots defeat the world's best human players at that game. [1]
Yan joined SenseTime, then one of China's most valuable computer vision startups, in 2015, beginning as an intern around the time he finished his doctorate. [1][3] He rose quickly through the company's research organization, eventually serving as a vice president, deputy dean of the SenseTime Research Institute, and chief technology officer of the Smart City business group. [1][3] In those roles he led the development of general-purpose computer vision models and the technical architecture behind SenseTime's smart-city products, and he became known for mentoring young researchers, including interns who were later recruited as Huawei "Genius Youth" hires. [3]
In December 2021 Yan left SenseTime to found MiniMax, which was formally incorporated in Shanghai in early 2022 under the legal entity Shanghai Xiyu Technology. [1][2] He co-founded the company with two fellow former SenseTime computer vision researchers, Yang Bin and Zhou Yucong, and took the roles of chairman, CEO, and CTO. [2] The company's name is a reference to the minimax algorithm from game theory and decision-making. [1]
Yan framed MiniMax from the outset as a bet on AGI rather than on any single product, and he has resisted the label of "China's OpenAI." His central technical thesis is that the essence of general intelligence is multimodal fusion, and that text, audio, image, and video capabilities must advance together, or the resulting "technical debt" will become a fatal weakness when those modalities are combined. [10] He has argued that AI is "not a mysterious black box" but an engineering problem that can be analyzed from first principles, and that "ordinary people plus scientific methods," rather than lone geniuses, can produce extraordinary results. [10]
MiniMax grew rapidly with backing from China's largest internet companies. According to its listing disclosures and press reports, it raised on the order of $850 million across four private rounds before its IPO. [7] A Series A of more than $250 million in mid-2023 was backed by Tencent, and a roughly $600 million Series B in March 2024, led by Alibaba, valued the company at about $2.5 billion. [2] Other investors included the game studio miHoYo, an early backer, along with Hillhouse, HongShan, IDG Capital, Gaorong Capital, and Yunqi Partners, and by 2025 the company's private valuation had reached roughly $4 billion. [2][5][7] MiniMax also drew unusually high-profile attention from outside China: Nvidia chief executive Jensen Huang publicly praised the company, and reports described a private meeting between Huang and Yan. [13]
MiniMax pursued consumer applications and foundation models in parallel. Its earliest consumer hit was an AI character-chat app, launched domestically as Glow in October 2022, that was pulled from Chinese app stores in 2023. [2] The company then split its companion product into Talkie for international markets, released in June 2023, and Xingye (星野) for China, released in September 2023; Talkie became one of the most widely used AI-native consumer apps worldwide. [2]
On the model side, MiniMax released the abab series of large language models, including the mixture-of-experts model abab 6.5 in April 2024, followed by the MiniMax-Text-01 and MiniMax-VL-01 foundation models in January 2025. [2] In June 2025 it released MiniMax-M1, the first in a family of open-weight reasoning models that the company iterated rapidly, through MiniMax-M2 and later versions such as M2.5 in February 2026, M2.7 in March 2026, and M3 in June 2026. [2] In multimedia, the Hailuo AI service (海螺, meaning "conch"), launched in March 2024, added the video-01 text-to-video model in September 2024 and later video, audio, and music models, while MiniMax Speech-02, released in April 2025, supported more than 30 languages. [2] The company also offers an enterprise and developer platform and an agent product, MiniMax Agent. [2]
By the time of its 2026 listing, MiniMax said its products reached more than 200 million users across over 200 countries and regions, with the majority of revenue coming from overseas markets, all built by a workforce of fewer than 400 employees. [4]
Around this period Yan publicly reflected on a strategic shift toward open source. "If we had to start over, we would have gone open source on day one," he said, arguing that building a technology brand was essential because "the greatest driver of this industry is technological evolution." [9] At the 2025 World Artificial Intelligence Conference in Shanghai he delivered a keynote titled "Everyone's AI," reiterating his view that AI would become steadily more accessible and that AGI would eventually be realized. [11]
MiniMax listed on the Hong Kong Stock Exchange on January 9, 2026. The offering, priced on January 8 at the top of its range at HK$165 per share, raised about $619 million (roughly HK$4.8 billion) and was heavily oversubscribed. [6][7] On its first trading day the stock roughly doubled, closing near HK$345, about 109 percent above the offer price. [5][12]
Reported valuations varied with timing and source. The IPO price implied a market capitalization of roughly $6.5 billion, up from a pre-listing private valuation of about $4 billion, and reports said the company's market value briefly surpassed HK$90 billion (more than $11 billion) during the first-day surge. [4][5] Yan's stake was valued at about $3.5 billion, making him a billionaire at 36 and, by several accounts, giving MiniMax one of the fastest paths from founding to public listing of any AI company, roughly four years. [1][4][8] The listing came during a wave of Chinese AI offerings: MiniMax is counted among the country's "AI tigers," also called the "six little dragons," a group that includes Zhipu AI, Moonshot AI (maker of Kimi), Baichuan, 01.AI, and StepFun, and Zhipu listed in Hong Kong around the same time. [5] In January 2026 Yan met China's Premier Li Qiang, becoming the second leader of a Chinese foundation-model company, after DeepSeek's Liang Wenfeng, to do so. [4]
| Year | Milestone |
|---|---|
| 1989 | Born in Xiayi County, Shangqiu, Henan |
| 2010 | Bachelor's degree in mathematics, Southeast University |
| 2015 | PhD in AI (computer vision), Institute of Automation, Chinese Academy of Sciences; joins SenseTime |
| 2015 to 2021 | Rises to vice president, deputy dean of research, and Smart City CTO at SenseTime |
| December 2021 | Leaves SenseTime and founds MiniMax with Yang Bin and Zhou Yucong |
| 2023 to 2024 | Talkie and Xingye companion apps; abab 6.5 LLM; Tencent- and Alibaba-led funding |
| June 2025 | Releases MiniMax-M1 reasoning model; open-source pivot |
| January 9, 2026 | MiniMax IPO on the Hong Kong Stock Exchange; Yan becomes a billionaire |