CtrlK
| Z.ai |
|---|
| 智谱 |
 |
| Type |
| Industry |
| Founded |
| Founders |
| Headquarters |
| Key people |
| Li Juanzi (Co-founder) |
| Liu Debing (Chairman) |
| Zhang Peng (CEO) |
| | Products | GLM (General Language Model) ChatGLM CodeGeeX AutoGLM Ying (text-to-video) CogVLM CogView CogVideoX MaaS Platform GLM-Z1 | | Revenue | US$42 million (2024)[1] US$168 million (2025, expected)[1]
| | Valuation | US$2.8 billion (December 2024)[2] | | Employees | ~2,000 (2025)[3] | | Website | / zhipuai.cn z.ai / zhipuai.cn |
Beijing Zhipu Huazhang Technology Co., Ltd.** (北京智谱华章科技有限公司), branded as Z.ai (formerly known as Zhipu AI until July 2025), is a Chinese artificial intelligence company specializing in large language models (LLMs) and generative AI technologies. Founded in 2019 as a spinoff from Tsinghua University, the company is one of China's "AI Tigers" and is considered the third-largest LLM market player in China's AI industry according to the International Data Corporation.[4]
Z.ai was founded on June 11, 2019, by Tsinghua University professors Tang Jie and Li Juanzi at the Tsinghua University Science Park in Beijing's Zhongguancun district.[5] The company originated from Tsinghua's Knowledge Engineering Group (KEG), where it was initially incubated.[5]
The founders faced early challenges securing initial investment. The administrative commission of Zhongguancun Science Park provided the team with three months of rent-free office space to help kickstart development.[6] Initially focused on building knowledge graphs, the company pivoted in 2020 to invest in developing large-model AI technologies, recognizing their potential before many competitors.[6]
In late 2020, Zhipu developed the GLM pre-training architecture. By 2021, they completed training of the GLM-10B model with tens of billions of parameters. In 2022, the company developed and open-sourced the GLM-130B, a massive bilingual pre-training model supporting Chinese and English.[1]
In 2023, Zhipu launched ChatGLM, a dialogue model with hundreds of billions of parameters, and upgraded it twice during the year. The open-source version, ChatGLM-6B, allowed developers to fine-tune and deploy the model locally.[1] The ChatGLM series has achieved over 30 million global downloads on platforms including Hugging Face.[7]
In March 2024, CEO Zhang Peng announced that Zhipu was developing Sora-like technology as a pathway to achieving artificial general intelligence (AGI).[8] In July 2024, the company debuted "Ying," a text-to-video model capable of generating six-second video clips.[4]
In October 2024, Zhipu released AutoGLM, an AI agent application that uses voice commands to complete tasks within smartphones, marking a significant advancement in "Phone Use" AI applications.[9]
On January 16, 2025, the United States Department of Commerce's Bureau of Industry and Security added Zhipu AI (Beijing Zhipu Huazhang Technology Co., Ltd.) and several affiliates to the U.S. Entity List, imposing additional export licensing requirements due to national security concerns.[10] The company rejected these allegations as baseless.[11]
In July 2025, Zhipu rebranded from "Zhipu AI" to "Z.ai" alongside the release of GLM-4.5 and GLM-4.5 Air, their next-generation language models.[12] The company announced its GLM-4.5 model would cost 11 cents per million input tokens and 28 cents per million output tokens, pricing it below competitors like DeepSeek.[12]
In September 2025, Z.ai offered migration plans for users of Anthropic's Claude AI to switch to GLM-4.5, capitalizing on global AI shifts.[13] CEO Zhang Peng stated publicly that full artificial superintelligence is unlikely by 2030, though systems may surpass humans in some aspects.[14]
Tang Jie - Co-founder and controlling shareholder. Tang is a professor at Tsinghua University's Department of Computer Science and Technology, Director of the Foundation Model Research Center at Tsinghua's AI Institute, and an IEEE, ACM, and AAAI Fellow. He directly owns 7.4081% of the company and controls approximately 17.3966% of voting rights through shareholding platforms.[11]
Li Juanzi - Co-founder and professor at Tsinghua University. She directs the university's Knowledge Engineering Group (KEG), which has incubated multiple AI startups including DeepLang and Moonshot AI.[5]
Liu Debing - Chairman and co-controlling shareholder with Tang Jie. He holds 0.2554% directly and controls significant voting rights.[11]
Zhang Peng - Chief Executive Officer. Previously served as deputy director of the Science and Technology Big Data Research Center at Tsinghua University's Institute of Data Science. Holds a doctorate in computer science from Tsinghua University.[15]
Zhang Fan - Former Chief Operating Officer, departed in 2025.[16]
The company has no single controlling shareholder. The effective controllers are co-founder Tang Jie and chairman Liu Debing, who together with CEO Zhang Peng and other aligned stakeholders collectively control 36.9647% of voting rights.[11]
The General Language Model (GLM) series is Z.ai's flagship line of pre-trained dialogue models, initially developed in collaboration with Tsinghua KEG in 2023.[4]
| Model | Release Date | Parameters | Key Features |
|---|---|---|---|
| GLM-130B | 2022 | 130B | Bilingual (Chinese/English) pre-training model[1] |
| ChatGLM-6B | 2023 | 6B | Open-source, locally deployable[7] |
| GLM-4 | January 2024 | Undisclosed | Longer contexts, multi-modal capabilities, "All Tools" capability[1] |
| GLM-4.5 | July 2025 | 355B total, 32B active | Mixture-of-Experts architecture, dual-mode system, 128K context[17] |
| GLM-4.5V | July 2025 | Not specified | Vision and multimodal support[17] |
| GLM-4.6 | September 2025 | Undisclosed | 200K context window, enhanced coding and reasoning, powers chat.z.ai[18] |
| GLM-Z1 | 2025 | Not specified | Reasoning-focused model[19] |
AutoGLM is an AI agent application launched in October 2024 that can simulate human operations on mobile phones and computers to perform various tasks.[20] Key features include:
Voice command control for complex multi-step tasks
Support for over 54-step operations across applications[21]
Integration with major Chinese apps including WeChat, Taobao, Meituan, and Xiaohongshu[22]
Cloud-based execution for privacy protection[23]
Partnership with Alibaba Cloud for smartphone AI agent deployment[24]
AutoGLM 2.0 was launched in August 2025, positioning it as the world's first mobile agent capable of handling tasks across different apps.[25]
CodeGeeX - Code generation tool supporting multiple programming languages[7]
Ying (Template:Lang-zh) - Text-to-video model launched July 2024, generates 6-second video clips[4]
CogVLM and CogView - Visual language models for image understanding and text-to-image generation[7]
CogVideoX - Text-to-video generation model[26]
Zhipu Qingyan (智谱清言) - All-modal AI assistant app with over 25 million users, generating annual recurring revenue exceeding 10 million yuan[1]
GLM-4-Voice - Speech model for voice interactions[27]
GLM-4-Flash - Free API model for real-time web retrieval and long-context tasks[28]
Z.ai has raised approximately $1.5 billion across 12 funding rounds since its founding:[29]
| Date | Round | Amount | Lead Investors | Valuation |
|---|---|---|---|---|
| September 2022 | Series B | Undisclosed | Qiming Venture Partners | Undisclosed |
| October 2023 | Series B+ | ~$350 million | Alibaba Group, Tencent | Undisclosed |
| May 2024 | Series C | $400 million | Prosperity7 Ventures (Saudi Aramco) | ~$3 billion[30] |
| December 2024 | Series D | $412 million | Various | $2.8 billion[2] |
| March 2025 | Series D+ | $257 million | Hangzhou Municipal Construction Investment, Huafa Group | ~$3 billion[31] |
| July 2025 | Strategic | $140 million | Pudong Venture Capital, Zhangjiang Group | >$2.79 billion[3] |
Major institutional investors include:[29]
Technology companies: Alibaba Group, Tencent, Ant Group, Meituan, Xiaomi
Investment firms: HongShan (formerly Sequoia China), Legend Capital, Qiming Venture Partners
State-backed funds: Beijing AI Industry Investment Fund, Zhuhai Huafa Group, Hangzhou Industrial Investment Group, Pudong Venture Capital, Zhangjiang Group
International: Prosperity7 Ventures (Saudi Arabia)
In March 2025, Z.ai filed for IPO counseling with China International Capital Corporation (CICC) as its sponsor, becoming the first of China's "AI Six" unicorns to pursue a public listing.[11] The company restructured from a foreign-invested limited liability company to a foreign-invested joint-stock company to facilitate the IPO process.[11]
Enterprise Services: MaaS (Model-as-a-Service) platform providing API access to GLM models
Consumer Applications: Zhipu Qingyan app with premium subscription services generating over 10 million yuan annually[2]
Partnerships: Collaborations with hardware manufacturers including Honor, Intel, and Huawei[5]
The company has adapted its models for domestic hardware, including Huawei Technologies' semiconductors, to broaden China's AI ecosystem amid U.S. export restrictions.[32]
In 2024, Z.ai's commercial revenue reached US$42 million, with expectations to quadruple to US$168 million in 2025.[1] The company's MaaS platform experienced a 30-fold year-on-year increase in revenue after significantly lowering API prices.[2]
Z.ai competes primarily with other Chinese AI companies including:
DeepSeek - Known for cost-efficient models
Baidu Ernie
Alibaba Tongyi Qianwen
Baichuan Intelligence
The company is considered one of China's "AI Tigers" alongside Moonshot AI, Baichuan, and MiniMax.[6]
Z.ai has established regional headquarters, subsidiaries, and research centers in multiple countries including the United States, United Kingdom, France, Singapore, and Malaysia.[1] The company has signed frontier AI safety commitments alongside global leaders including OpenAI, Google, Microsoft, and Meta.[1]
In 2025, the company formed a coalition with representatives from 10 Belt & Road and ASEAN countries to help member states develop sovereign AI infrastructure.[33]
Z.ai has formed strategic partnerships across various industries:
BioGeometry (February 2024) - Building large multimodal models for life sciences and medical research[34]
BYOND ASIA - Developing hyper-realistic digital humans for the entertainment industry[35]
Alibaba Cloud - Deployment of AutoGLM AI agents for smartphone users[24]
Huawei Technologies - Adaptation of models for domestic semiconductors[32]
Z.ai's founding team and researchers have made significant contributions to AI research, with publications including:
"ChatGLM: A Family of Large Language Models from GLM-130B to GLM-4 All Tools" (2024)[7]
"CodeGeeX: A Pre-Trained Model for Code Generation" (KDD 2023)[36]
Multiple papers on large language models, knowledge graphs, and AI agents
Mixture-of-Experts (MoE) Architecture: GLM-4.5 uses 355 billion total parameters with only 32 billion active per inference[17]
Dual-mode System: "Thinking mode" for complex reasoning and "non-thinking mode" for instant responses[37]
Agentic AI Capabilities: Integration of reasoning, coding, and tool use in single models[37]
Long-context Processing: Support for up to 200,000 tokens in GLM-4.6[18]
CEO Zhang Peng has stated the company's mission is to achieve "super cognitive intelligence beyond human level."[5] An internal saying at the company states: "No matter how much money we raise or how much money we make, it will be a hindrance on our road to AGI."[6] Zhang has also publicly commented on AI timelines, stating in September 2025 that full artificial superintelligence is unlikely by 2030 but that artificial general intelligence could emerge in limited domains by then.[14]
Artificial intelligence in China
Tsinghua University
Baidu
Entity List