Zhipu AI Overtakes MiniMax in Market Valuation Race

🔑 Enhanced Key Takeaways

•Zhipu AI debuted on the Hong Kong Stock Exchange on January 8, 2026, raising approximately $558 million at an initial valuation of around $6.57 billion. Its shares have since surged by nearly 1,600%, pushing its market capitalization above $80 billion by late May 2026.
•MiniMax also went public in Hong Kong on January 9, 2026, raising about $619 million with an initial valuation of approximately $6.5 billion. Its shares have risen more than fourfold since its debut, reaching a market value of roughly $33.6 billion by late May 2026.
•Both Zhipu AI and MiniMax are pursuing secondary listings on the A-share market in mainland China, with MiniMax formally initiating its IPO tutoring process on the Shanghai Stock Exchange's STAR Market in May 2026, and Zhipu AI appointing advisors for a similar listing in February 2026.
•Chinese AI models, including MiniMax's M2.5, Moonshot AI's Kimi K2.5, and Zhipu AI's GLM-5, collectively accounted for 61% of total token consumption among the top ten models on OpenRouter, the world's largest LLM API aggregation platform, by February 2026.
•Zhipu AI is actively expanding its global footprint, particularly in Southeast Asia, and is involved in building 'sovereign AI' infrastructure for partner countries through initiatives like the 'International Alliance for Independent Large Model Co-construction'.

📊 Competitor Analysis▸ Show

Feature/Category	Zhipu AI (Knowledge Atlas Technology)	MiniMax	Other Key Chinese Competitors
Primary Focus	General-purpose AI foundation models (GLM series), coding, reasoning, vision, agentic abilities, enterprise solutions.	Multimodal AI models (text, audio, image, video, music), consumer applications (Talkie, Hailuo AI), enterprise API services, coding, agentic workflows.	Moonshot AI (Kimi series, ultra-long context), DeepSeek (high-performance open-source models, cost-efficient), Baidu (Ernie Bot, search-data integration), Alibaba (Qwen family, multilingual, cloud scale).
Flagship Models	GLM-5 (744B MoE), GLM-4.7, GLM-4.5.	MiniMax M3, MiniMax M2.5, Hailuo AI, Speech 2.8, Music 2.6.	Kimi (Moonshot AI), DeepSeek-V2, DeepSeek-Coder (DeepSeek), Ernie Bot (Baidu), Qwen (Alibaba).
IPO/Valuation (Jan 2026)	HKEX IPO on Jan 8, 2026. Raised ~$558M. Initial valuation ~$6.57B. Market cap >$80B by late May 2026.	HKEX IPO on Jan 9, 2026. Raised ~$619M. Initial valuation ~$6.5B. Market cap ~$33.6B by late May 2026.	Moonshot AI reportedly seeking up to $2B at $30B valuation (June 2026).
Pricing Strategy	Aggressive pricing for AI services, e.g., AI coding tool at <$3/month, aiming to trigger global AI price war. GLM-4.5-Air input API cost 0.8 RMB/million tokens.	MiniMax M2 costs 92% less than Claude Sonnet 4.5, with competitive API pricing.	DeepSeek's V4 model (April 2026) offered at a fraction of competitors' costs, driving model democratization.
Key Benchmarks	GLM-4.7 achieved 73.8% on SWE-bench Verified. GLM-5 scored 77.8% on SWE-bench Verified, rivaling Claude Opus 4.5 and GPT-5.2.	MiniMax M3 claims to outperform GPT-5.5 and Gemini 3.1 Pro on SWE-Bench Pro and approaches Opus 4.7. MiniMax M2 achieved 69.4 on SWE-bench Verified.	DeepSeek-V2 and DeepSeek-Coder achieved state-of-the-art results on coding and reasoning benchmarks.
Global Market Share	GLM-5 from Zhipu AI was among the top three most-used models globally on OpenRouter by February 2026.	MiniMax M2.5 was among the top three most-used models globally on OpenRouter by February 2026.	Kimi K2.5 from Moonshot AI was among the top three most-used models globally on OpenRouter by February 2026.
International Strategy	Focus on Southeast Asia, building 'sovereign AI' infrastructure, leading 'International Alliance for Independent Large Model Co-construction'.	Serves over 1 million enterprise and developer customers globally, with a global user base of approximately 300 million.	Chinese models are pursuing a strategy of wide diffusion and cheap tokens to gain market share globally.

🛠️ Technical Deep Dive

Zhipu AI GLM Models:
- GLM-4.5: An open-source large language model built with a Mixture-of-Experts (MoE) architecture, featuring 355 billion total parameters and 32 billion active parameters. It includes a 'thinking mode' for dynamic response based on task complexity and supports a 128K context window. It also uses 3D Rotated Positional Encoding (3D-RoPE) for vision capabilities.
- GLM-4.7: Released December 2025, built upon approximately 400 billion parameters, supporting a 200,000-token context window with a maximum output capacity of 128,000 tokens. It has an inference efficiency of 55 tokens per second.
- GLM-5: Released February 2026, a 744-billion parameter Mixture-of-Experts language model that activates only 40 billion parameters per token, utilizing a 256-expert MoE with efficient routing. It supports a 200K+ context window and includes native tool calling and function execution.
MiniMax Models:
- Multimodal Foundation Models: MiniMax develops models capable of understanding, generating, and integrating text, audio, image, video, and music.
- MiniMax M1: Utilizes a Mixture-of-Experts (MoE) architecture with 456 billion total parameters, where a sparse subset of approximately 45.9 billion parameters (10% of total) is activated during inference. It incorporates a custom 'lightning attention' mechanism and supports up to a 1 million token context window.
- MiniMax M2: A compact, high-efficiency large language model with 230 billion total parameters and 10 billion activated parameters, optimized for end-to-end coding and agentic workflows. It has native support for tools like Shell, Browser, and Python interpreter.
- MiniMax M3: Features a novel attention architecture called MiniMax Sparse Attention (MSA) and supports ultra-long context windows of up to 1 million tokens. It is natively multimodal, supporting image and video input, and can operate a desktop computer.

🔮 Future ImplicationsAI analysis grounded in cited sources

Chinese AI companies will intensify global competition through aggressive pricing and open-source strategies.

Zhipu AI is already pushing affordable AI services globally, and Chinese models are increasingly dominating global token consumption due to low-cost and open-source approaches, signaling a structural shift in the global AI market.

The Chinese AI market will see further consolidation and dual listings among its leading players.

Both Zhipu AI and MiniMax are pursuing secondary listings on the Shanghai Stock Exchange's STAR Market, indicating a trend for major players to secure more capital and market presence across multiple exchanges.

⏳ Timeline

2019

Zhipu AI founded by Tsinghua University professors Tang Jie and Li Juanzi.

2023

Zhipu AI raised 2.5 billion yuan (approx. $350 million USD) from investors including Alibaba Group and Tencent.

2025-12-22

Zhipu AI released its GLM-4.7 large language model.

2026-01-08

Zhipu AI listed on the Hong Kong Stock Exchange, raising over HK$4.3 billion (approx. $558 million) at an issue price of HK$116.2 per share.

2026-02

Zhipu AI released its GLM-5 large language model and appointed advisors for a secondary listing on Shanghai's STAR Market.

2026-05

Zhipu AI's market capitalization surpassed $80 billion, following a nearly 1,600% surge in its shares since its IPO.

Zhipu AI Overtakes MiniMax in Market Valuation Race

⚡ 30-Second TL;DR

Key Points

🧠 Deep Insight

🔑 Enhanced Key Takeaways

🛠️ Technical Deep Dive

🔮 Future ImplicationsAI analysis grounded in cited sources

⏳ Timeline

📎 Sources (31)

👉Related Updates

Systemic mismatch in the current job market

Chinese robot maker AgiBot pursues Hong Kong IPO