China's generative AI landscape reached a historic milestone in January 2026, when two of the country's leading “AI tigers” – Beijing-based Zhipu AI and Shanghai-based MiniMax – successfully completed initial public offerings on the Hong Kong Stock Exchange (HKEX). These debuts not only provide the capital necessary for the global AI arms race, but also highlight two distinct philosophies for achieving artificial general intelligence (AGI).
The divergence between these two organizations begins with their institutional legacies. Zhipu AI is a direct derivative of the Knowledge Engineering Group (KEG) at Tsinghua University. Led by professors Tang Jie and Li Juanzi, the company is research-intensive, prioritizing core artificial intelligence infrastructure and knowledge engineering. Zhipu positions itself as a “national core platform” provider focusing on on-premises and cloud deployments for government and enterprise sectors, including finance, healthcare and manufacturing.
MiniMax, on the other hand, represents the entrepreneurial spirit of the Shanghai technology cluster. Founded by computer vision veterans from SenseTime, most notably founder Yan Junjie, the company adopts an industrial, product-centric ethos. While Zhipu builds the foundation, MiniMax focuses on the application layer, developing multimodal consumer-facing products.
Both companies' technical strategies focus on the Mixture of Experts (MoE) architecture, but prioritize different dimensions of performance.
- Zhipu AI – GLM-4.7: This flagship model contains 358 billion parameters (with 32 billion activated per token) and focuses on high-fidelity reasoning. A key innovation is the “Deep Thinking” mode, which enables multi-step planning and autonomous debugging – capabilities that allow you to solve complex software engineering tasks with a 73.8% success rate on the SWE platform-verified leaderboard. Its AutoGLM agent further reflects this “action-oriented” research by autonomously performing tasks on mobile devices via voice and visual prompts.
- MiniMax – M2.1: The company focused on the computational limitations of traditional transformers through its hybrid “Lightning Attention” architecture. This allows the M2.1 to support an unprecedented native context window of 4 million tokens – equivalent to approximately 60 books. By enabling agents to “remember” entire codebases or huge sets of documents in a single line, MiniMax pioneers the “RAG Killer” paradigm, potentially eliminating the need for complex extended search generation pipelines in many enterprise use cases. MiniMax products, including text-to-video (Hailuo AI) and multilingual speech models, are aimed directly at developers and end users, with an emphasis on practicality and scalability.
Following Zhipu AI's addition to the US Entity List in January 2025, it remains primarily focused on the domestic market. The designation limits Zhipu's access to high-end hardware and software in the U.S., forcing it to focus on building a self-sustaining domestic artificial intelligence stack for Chinese institutional clients.
Unencumbered by such restrictions, MiniMax is pursuing an aggressive international strategy. With almost 73% of its revenue generated by the end of 2025 in overseas markets, MiniMax has successfully engaged global developers through roadshows in cities such as San Francisco and New York. Its consumer social media app Talkie has over 32.8 million monthly active users, positioning MiniMax as a direct global competitor to Western labs like Character.AI.
From 2026, Zhipu AI and MiniMax will be the two pillars of China's generative AI ambitions. Zhipu AI provides the technical and institutional “deep tech” foundations essential to national sovereignty, while MiniMax demonstrates the country's ability to create world-class, cost-effective consumer applications. Their simultaneous public debuts signal that China's AI sector has moved beyond the experimental stage and into a mature era of commercial validation and global competition.
