Z. ai

• 1. overview (756 chars)

  Z. ai (website: https://z. ai) is an AI model experience and creative content generation platform officially launched by Zhipu AI on April 15, 2025. It uses the new domain name z. ai as a global unified AI interaction portal. The platform integrates the latest GLM series of large models from Zhipu. It is completely free and open to users around the world and can be used without logging in. Z. ai is positioned as a "free and open source alternative to ChatGPT" and supports core functions such as intelligent dialogue, code generation, content creation, Vincent diagrams, PPT production, and long text processing.   On January 8, 2026, Zhipu AI was listed on the main board of the Hong Kong Stock Exchange (stock code 02513. HK), becoming the "first large-scale model stock in the world." The Z. ai platform, as its consumer-grade entrance, carries Zhipu's mission of "Inspiring AGI to Benefit Humanity".

• 2. Product background (1250 chars)

  Zhipu AI (full name: Beijing Zhipu Huazhang Technology Co., Ltd., Zhipu AI) was established in June 2019. Its core technology comes from the Knowledge Engineering Laboratory of Tsinghua University. It is one of the largest independent large model companies in China. The founder and CEO is Liu Debing, and Tang Jie serves as the chief scientist.   Zhipu’s cumulative financing exceeds 16 billion yuan, with investors including Hillhouse Capital, Qiming Venture Partners, Meituan, Alibaba, Tencent, Xiaomi, and state-owned assets in Chengdu, Hangzhou, and Zhuhai. In December 2024, the valuation exceeded 20 billion yuan and was regarded as the first of the "Six Little Dragons of Large Models". It submitted listing guidance filing in April 2025 and was listed on the main board of the Hong Kong Stock Exchange on January 8, 2026. The issue price was HK$116.20 per share, the opening price on the first day was HK$120 per share, and the market value was HK$52.828 billion. The Hong Kong public offering was subscribed 1159.46 times.   In terms of financial data, Zhipu’s revenue will be 57.4 million yuan in 2022, 124.5 million yuan in 2023 (+117% year-on-year), 312.4 million yuan in 2024 (+151% year-on-year), and 191 million yuan in the first half of 2025 (+130% year-on-year). However, the cumulative net loss in three and a half years has reached 6.248 billion yuan. Every 1 yuan of revenue corresponds to about 8 yuan of R&D investment, and the pressure for commercialization is not small.

• 3. Product functions and usage experience (1226 chars)

  The Z. ai platform integrates 8 free AI models, covering three core capabilities: base, reasoning, and contemplation. The base model GLM-4.7-32B is responsible for general dialogue and copywriting creation; the inference model GLM-Z1-32B-0414 has a measured inference speed of 200 Tokens/second, setting the fastest domestic business model record; the contemplation model GLM-Z1-Rumination focuses on in-depth research and iterative optimization.   Core functions include: intelligent dialogue and multi-round dialogue, supporting deep reasoning on complex problems; code generation and full-stack development, covering complete project delivery from front-end to back-end; content creation, covering advertising copy, social media posts, blog posts, product descriptions, etc.; Vincent diagrams, supporting the creation of multiple visual styles; PPT production and presentation slide generation; long text processing, with a context window of up to 200K tokens; document analysis and data extraction.   The GLM-5.1 flagship model released in April 2026 has greatly enhanced coding capabilities. It can continuously work autonomously for up to 8 hours on a single task, completing complete project delivery from planning, execution to iterative optimization, and ranking first in the world on authoritative programming lists such as SWE-Bench Pro. The Z. ai platform has gradually integrated these latest flagship models.

• 4. Pricing and business model (542 chars)

  The Z. ai platform itself is completely free, has unlimited usage, and can be accessed without logging in.   The platform achieves commercialization through the following paths: API call service (MaaS platform) charges for developers, Coding Plan paid subscriptions for professional programming users, and privatized deployment and customization services for enterprises. The open source model follows the MIT license, allowing free commercial use and secondary development. Compared with international competing products, the GLM series API price is about 1/30 of DeepSeek-R1, which is far lower than the charging standard of GPT-5.

• 5. User reviews (748 chars)

  Z.ai’s free policy is widely welcomed by users and has significantly lowered the threshold for domestic users to use AI. After the platform was launched in April 2025, a large number of tutorial articles appeared in Zhihu and other communities, teaching users how to use the three free models.   User feedback shows that the platform has fast response speed, smooth interface operation, and programming capabilities that are close to the top international standards. The understanding and optimization of Chinese scenes have become a differentiation bonus.   In terms of pain points, the Coding Plan subscription service is often out of stock due to difficulties in buying it. Some users need time to adapt to the change of z. ai brand domain name. Zhipu's business model that continues to lose money has also caused some users to worry about the long-term operational stability of the platform.

• 6. Industry evaluation (542 chars)

  Zhipu AI's positioning in the field of domestic and foreign AI has shifted from "catcher" to "competitor". GLM-4.5 ranked third in the world in 12 authoritative international benchmark tests, setting a new historical record for domestically produced large models; GLM-5's programming capabilities reached the top of the world; GLM-5.1 continues to consolidate this advantage.   In the list of the world's ten most influential AI companies published by Time magazine in April 2026, Zhipu was selected together with ByteDance, OpenAI, Google, Meta, Anthropic, Alibaba, etc., becoming the only domestic large-scale model company selected.

• 7. Controversies and Risks (450 chars)

  There are three main challenges facing Zhipu: continuous losses (accumulated net loss of 6.248 billion yuan in three and a half years), intensifying competition (big companies such as Baidu Wenxin, Alibaba Tongyi, and Byte Doubao are accelerating their entry into the market), and high valuation pressure brought by internationalization. In addition, in 2025, media reported that 35 apps including Zhipu Qingyan were notified for "violating the collection of personal information" and compliance rectification is in progress.

• 8. Applicable groups and usage suggestions (532 chars)

  Z. ai is suitable for the following users: Users who need a free AI assistant for daily conversations and information queries can directly use the base model; for the developer group, Z. ai provides nearly cutting-edge programming capabilities for free and is a cost-effective alternative to GitHub Copilot; for content creators, copywriting and social media operations can use the contemplation model for in-depth research and iterative optimization; for academic researchers, the 200K context supports long papers and complex document analysis; for the student group, free access to math tutoring and homework assistance.

No in-depth report yet

• 

  JVasquez_Pro

  —

  GLM-5.1编程能力太顶了，SWE- Bench直接登顶全球第一，关键是还免费！

---

• 

  h1uy3ot

  —

  说实话智谱这一波真的很良心，GLM-4.7、GLM-5、GLM-5.1全部免费，随便用

---

• 

  莲花_9

  —

  用Z. ai写代码比用Claude还顺手，Tab补全很智能

---

• 

  万珍

  —

  智谱这次开源的GLM-5.1确实牛，直接对标GPT-5和Claude Opus

---

• 

  PEpar

  —

  港股上市后第一天就涨了11%，市值突破600亿港元，厉害了

---

• 

  LUphi

  —

  每天用Z. ai写文案，免费的谁不爱啊

---

• 

  CLee_Pro8

  —

  全球大模型第一股，智谱这个名号真的响

---

• 

  qj_5df

  —

  智谱的AutoGLM沉思智能体太方便了，浏览器操作全自动

---

• 

  DTaylorII

  —

  GLM-5.1可以连续工作8小时，交付工程级成果，真的绝了

---

• 

  EricYoung_66186

  —

  Z. ai网页直接生成预览的功能太实用了，做前端的我爱了

---

• 

  吕强哲

  —

  用Z. ai复刻了B站网友的创意，效果真的惊艳

---

• 

  Daniel_SmithK31

  —

  智谱中文理解确实比GPT强，用起来很舒服

---

• 

  REric_io

  —

  GLM-5的推理速度达200 Tokens/秒，国产之光！

---

• 

  Alexander.Ward_Plus

  —

  三大模型免费开放，智谱这波格局打开

---

• 

  Gabriel.Walker_77

  —

  清华系的技术就是扎实，模型矩阵很全

---

• 

  mi9cbtfpxc_lab

  —

  《时代》十大最具影响力AI公司，智谱入选了！

---

• 

  VBaileyZ0

  —

  Coding Plan国际版也能用，虽然经常缺货但比没有强

---

• 

  Donna_Rogers_20221

  —

  Z. ai界面很简洁，比智谱清言更好上手

---

• 

  Donald.Long_Pro

  —

  GLM-5编程能力能和GPT-5.3- Codex同台竞技了

---

• 

  Nathan_Mitchell_Pro70

  —

  不过智谱亏损还挺严重的，商业化还是要加快啊

No reviews yet