The company priced its shares at 165 Hong Kong dollars (about $21) each and saw them close more than double that at 345 Hong Kong dollars (about $44), the Financial Times reported Friday.
The IPO brought MiniMax’s market capitalization to $13.5 billion, according to the report.
MiniMax announced the listing in a Friday post on LinkedIn, adding: “This isn’t the finish line. It’s fuel for our next leap toward AGI [artificial general intelligence].”
“Built on one belief: Intelligence with Everyone,” the company said in the post. “We believe advanced AI should be accessible and beneficial to a wide range of users and industries. This guiding principle underscores everything we do: from research and development to how we engage with our users and partners.”
According to the FT report, another Chinese AI company, Zhipu, raised $558 million in its Hong Kong IPO on Thursday (Jan. 8) and has seen its shares rise 37% since then.
Both Zhipu and MiniMax could use their IPO revenues to expand overseas, the report said.
MiniMax is known for its multimodal models with image and video generation capabilities, and it generates most of its revenue from consumer applications such as the character chatbot app Talkie and the video generation platform Hailuo AI, per the report.
PYMNTS reported in March that Chinese AI startups came into focus after DeepSeek took Silicon Valley by surprise with foundation models that performed at par with the best AI models but at a fraction of the price.
Reuters reported Dec. 23, 2025, that global investors are betting on Chinese AI companies in hopes of finding the next DeepSeek and to diversify their investments.
Meanwhile, The Information reported Friday that DeepSeek is expected to launch its next-generation AI model, V4, within weeks. The report said that the model has strong coding capabilities, made breakthroughs in handling and processing long coding prompts, and made improvements to its ability to understand data patterns through every stage of the training process.
PYMNTS reported Monday (Jan. 5) that DeepSeek said it has developed a new way to train large language models (LLMs) that allows performance to improve without a proportional increase in training costs.
