zhipu

On September 2, the domestic large model manufacturer Zhipu opened an official flagship store on Tmall, selling products such as the Coding Plan package.Currently available products include personal versions like Lite, Pro, Max, etc., which can be purchased via monthly/quarterly/annual subscriptions. In the future, buying Tokens will also be possible on the e-commerce platform.

Robonomics author FD published the China Open-Source LLM Tracker on September 1, 2026, stating that the total ARR of China's open-source large models is approximately $10-15 billion, with core LLM revenue around $6-8 billion after excluding ByteDance / Seedance. Growth has not relied on price wars; after DeepSeek raised prices by about 3-12 times, usage still increased, and the average price of the Zhipu API rose by 101% while token usage exceeded 40 times within the year. The best estimate for overseas revenue is approximately $2 billion.Zhipu MaaS ARR increased from about $250 million in March to an annualized monthly rate of about $1.6 billion in August, with a weekly annualized rate of $2 billion. August revenue has already surpassed its API revenue for the first half of the year; the API gross margin is 24.6%, and inference costs per token decreased by 80%. MiniMax ARR rose from about $100 million in December 2025 to over $800 million in annualized weekly revenue by August 2026, with B2B accounting for about 80%. Kimi ARR increased from about $100 million in March to about $300 million by mid-June. DeepSeek's revenue from January to July was approximately $70 million, with the latest ARR estimated at about $1.2 billion.ByteDance's annualized AI revenue is approximately $4 billion, of which Seedance accounts for about $2 billion.

According to Gate market data, as of the close of the Hong Kong stock market, the Hong Kong Hang Seng Index rose by 0.07%, while the Hang Seng Tech Index fell by 0.33%. Domestic property stocks led the gains, with Sunac China rising over 10%, CIFI Holdings Group rising over 5%, and Country Garden and Vanke Enterprises rising over 2%.The non-ferrous metals sector rose, with China Nonferrous Mining rising over 4%, Tianqi Lithium rising over 3%, and China Aluminum rising over 3%. The PCB and storage concepts weakened, with Shenghong Technology falling nearly 13%, Chipbond Technology falling over 8%, and Zhaoyi Innovation and Lattice Semiconductor falling over 5%.Large model concept stocks declined, with Zhiyu falling over 6% and MINIMAX-W falling over 4%.

According to the Securities Times, on August 27, the reporter learned from SenseTime that SenseTime's large device provides domestic computing power support and Token services for the GLM-5.3-Flash recently launched and open-sourced by Zhipu, which is considered another benchmark for the large-scale commercial use of domestic computing power. It was revealed that the daily average Token service volume of SenseTime's large device reached 24.2 trillion in July, and it is expected to exceed a daily average of 100 trillion by the end of 2026.

On August 27, the one-stop AGI infrastructure platform B.AI announced that the Zhipu GLM-5.3-Flash (320B-A18B) officially launched on the platform and is fully open for free in the API official group. As the first native multimodal model in the GLM-5 series, GLM-5.3-Flash features a total of 320 billion parameters and 18 billion active parameters, supported by computing power from purely domestic AI chips. It introduces a mixed architecture of sparse and linear attention for the first time, combined with a 1M ultra-long context, offering both powerful performance and exceptional cost-effectiveness. Starting today, developers can call it at zero cost through the B.AI API official group, easily handling complex reasoning, long text processing, and multimodal tasks.

According to Gate market data, the Hong Kong stock market closed with the Hang Seng Index down 0.34% and the Hang Seng Tech Index down 0.13%; Zhipu (02513.HK) rose 12.62% for the day.

According to Gate market data, the Hong Kong stock market opened, with the Hang Seng Index rising 0.48% and the Hang Seng Tech Index rising 0.85%. Zhipu (02513.HK) opened up over 8%, recently launching a lightweight flagship model benchmarked against DeepSeek.

Zhipu released the lightweight flagship model GLM-5.3 Flash, aimed at competing with DeepSeek V4 Flash, and utilizes over 100,000 domestic chip clusters to provide inference services. The model has a parameter scale of approximately 300B, about 40% of GLM-5.3, and is roughly on par with DeepSeek V4 Flash; it adopts the MoE architecture, with actual activated parameters around 18B, possesses native multimodal capabilities, supports image and video input, and is the first new model with multimodal capabilities following Zhipu's focus on coding strategy.The comprehensive evaluation score exceeds that of the previous flagship GLM-5.2, is on par with Claude Opus 4.8, and is higher than the relevant scores of DeepSeek V4 Pro official version. The pricing is 0.8 yuan per million tokens for input, 2.8 yuan for output, and 0.23 yuan for cache hits, approximately one-tenth of GLM-5.3, with a half-price discount offered for the first two weeks after launch. The model is fully compatible with domestic computing power, online traffic is handled by domestic chips, and Zhipu claims that the cluster hardware efficiency and unit token cost have reached levels comparable to mainstream NVIDIA GPUs; this model was previously tested anonymously on overseas platforms under the code name Ox Alpha.

Zhipu officially announced the launch and open-sourcing of GLM-5.3-Flash, which is the first native multimodal model in the GLM-5 series.It is reported that the overall performance of GLM-5.3-Flash exceeds that of GLM-5.2, with programming and Agent evaluations approaching Claude Opus 4.8, but at only one-tenth the price of GLM-5.2. It also features a new foundational model, introducing a mixed architecture of sparse attention and linear attention for the first time in the main GLM series, and is pre-trained with 30T Token multimodal data.Zhipu stated that to gather extensive and professional feedback from a wide range of users, large-scale testing was conducted with the anonymous model Ox-Alpha (referred to as "Niu Lai" in the Chinese community) on OpenCode and OpenRouter before the official release. Ox-Alpha quickly became the most popular model of the week, setting a new high for call volume on both platforms, with all request traffic supported by domestic chip computing power.Previously, the community's DeepSWE small sample test for Ox Alpha once achieved 80%, but that was based on only 10 questions. After expanding the sample, the score fell back to about 63%, and testers also actively corrected the initial claim. This score still belongs to the top tier, but is not as exaggerated as the initial 80%. After the weights are open-sourced, developers can directly deploy using frameworks like vLLM, SGLang, KTransformers, without needing to go through the anonymous model's API.

According to Bloomberg, Chinese artificial intelligence company Z.AI (Zhìpǔ) has confirmed that the mysterious artificial intelligence model Ox Alpha, which has rapidly risen to the top of online usage rankings, is a new iteration of its GLM series.The company stated in response to a Bloomberg News inquiry that it will release the model weights for Ox Alpha tonight.

According to Bloomberg, the Chinese AI model company Z.AI (智谱, 2513:HK) listed in Hong Kong and MiniMax Group (100:HK) are facing record short-selling bets. Data shows that MiniMax's short position has accounted for about 20% of its free float, while Z.AI is around 6%. These positions have risen to high levels ahead of the two companies' upcoming earnings reports, reflecting investors' increasing concerns about the intense competition in the Chinese AI sector.After both companies completed high-profile listings earlier this year, their stock prices have retreated from historical highs, but short-selling activity has further intensified. Market attention is focused on the escalating competition among domestic large model manufacturers and the resulting pressure on profit prospects and valuations.

Z.ai released GLM-5.3, which is based on the same foundational model as GLM-5.2, achieving capability improvements through expanded post-training. The official statement claims that GLM-5.3 has improved by 50% on the internal Z.ai Code Bench coding benchmark compared to GLM-5.2, and has reached leading levels among open models in public benchmarks such as Terminal Bench 3 and Agents' Last Exam.In terms of cybersecurity, GLM-5.3 achieved a score of 84.5% in the CyberGym vulnerability discovery test, and showed significant improvements in exploitation chain-related tests such as ExploitBench and ExploitGym compared to its predecessor.

The number of registered users on the Zhipu MaaS open platform (i.e., API users) has approached 7 million, an increase of about 2 million compared to early July, including 23,000 enterprise customers. The user count for the developer product ZCode, which competes with Codex, surpassed one million within one month of its launch.Since the beginning of this year, Zhipu's ARR has grown 15 times. It is understood that after the release of new models by Kimi and Alibaba, Zhipu's ARR growth rate has not been affected. Some investors have revealed that the current ARR has reached $2 billion, but Zhipu's official stance has denied this, with people close to the company indicating that the actual figure should be higher. On July 21, it was reported that Zhipu has built a domestic AI computing power infrastructure with a scale of 1GW. It is understood that more than 50,000 domestic computing chips have been activated to alleviate the growing inference demand. On July 31, Zhipu's long-term Coding Plan, which had significantly increased in price, was opened for purchase.

According to Gate market data, the Hang Seng Index turned to decline in the afternoon, and the Hang Seng Tech Index's increase narrowed to 0.44%. Zhizhu (02513.HK) fell by 8%, while SMIC (00981.HK) and Li Auto (02015.HK) both dropped by more than 2%.

According to official news, the Gate gStocks section will launch ZHIPUG (Zhipu AI), MINIMAXG (MiniMax), TENCENTG (Tencent Holdings), HOODG (Robinhood), IBMG (IBM), AMZNG (Amazon), BEG (Bloom Energy), DELLG (Dell Technologies), FLNCG (Fluence Energy), SMHG (VanEck Semiconductor ETF), and SOXSG (Direxion Semiconductor 3x Short ETF) for spot trading of 11 gStocks tokenized securities on July 30 at 14:00 (UTC+8). The gStocks tokenized securities products have a 1:1 full native reserve, providing a one-stop solution for 24/7 securities trading.In addition, Gate will launch the fourth phase of the gStocks new coin trading carnival from July 30 at 16:00 to August 6 at 16:00 (UTC+8). Completing designated gStocks trades will allow participants to receive newcomer rewards, participate in the holding plan and trading competition, sharing a total prize pool of $50,000 SNDKG.

According to Gate market data, the Hong Kong Hang Seng Tech Index rose over 1% during the session. Zhihu (02513.HK) rose over 8%, Xiaomi Group (01810.HK) rose over 5%, Trip.com Group (09961.HK) rose nearly 4%, and Meituan (03690.HK) rose over 3%.

According to the official announcement, Bitget has launched Quanto perpetual contracts for Tencent (TENCENT), Zhipu AI (ZHIPU), and Xiaomi (XIAOMI), supporting up to 20 times leverage and 24/7 trading.TradFi Quanto perpetual contracts allow users to trade Hong Kong stocks and other overseas stocks priced in currencies other than USD directly using USDT, without the need to exchange foreign currency throughout the process. The contracts are quoted in Hong Kong dollars, while the margin deposit and profit and loss settlement are completed in USDT, effectively avoiding additional risks brought by exchange rate fluctuations while simplifying the trading process.

According to market news, Goldman Sachs reported on July 18 that the Dark Side of the Moon released the Kimi K3 model, with 28 trillion parameters, surpassing Claude Fable 5 and GPT-5.6 Sol on the Arena.ai programming leaderboard, reaching a new high for Chinese models with an API pricing of $2.3 per million tokens. Two days after the release, Zhiyu fell by 28%, MiniMax dropped by 16%, the Nasdaq 100 index futures fell by more than 1.8%, and the Philadelphia Semiconductor Index has cumulatively dropped by more than 18% from its peak.Goldman Sachs believes that Kimi K3 marks a turning point: a Chinese laboratory that cannot match the largest pre-trained computing power in the West has rapidly narrowed the gap with top American models through architectural innovation and reinforcement learning, proving that "scaling" is no longer the only winning path. Goldman Sachs warns that the "era of computing power expansion" may be coming to an end, and the investment logic for AI infrastructure built around "the more computing power, the better" needs to be rewritten. Goldman Sachs maintains a buy rating on MiniMax and a neutral rating on Zhiyu. Future attention will focus on the intensive release of 20-50 trillion parameter models such as Zhiyu GLM, Alibaba Qwen, and MiniMax M3 Pro.

According to "AI Technology Review," China's leading large model company Zhipu has invested hundreds of millions of yuan to acquire the AI heterogeneous computing power software infrastructure company Zhongke Jiahe. This move aims to completely address Zhipu's shortcomings in the underlying engineering and compiler capabilities of large models, in response to the structural shortage of computing power and high-concurrency inference challenges caused by a surge in user numbers.Zhongke Jiahe's technology originates from the Compiler Laboratory of the Institute of Computing Technology, Chinese Academy of Sciences, founded by Dr. Cui Huimin. Its core team has been deeply involved in the development of compilers for several domestic chips, including Loongson, Sunway, Cambricon, and Huawei Ascend. Zhongke Jiahe's core advantage lies in its virtual instruction set technology, which can unify different brands and models of chip ecosystems through middleware software, assembling scattered domestic chips into a unified ultra-large-scale cluster, thereby significantly improving overall computing power utilization; its SigInfer inference engine is claimed by the official source to reduce the inference latency of large models by up to 74 times.Recently, Zhipu's Coding Agent business has experienced explosive growth. The newly released GLM-5.2 large model saw a 27-fold increase in daily Token call volume during its first week on the aggregation platform, leading to the exposure of systemic engineering bottlenecks in its inference infrastructure under high concurrency and long context scenarios. After being placed on the U.S. Entity List, Zhipu has actively promoted domestic alternatives and has now completed inference adaptation for eight major domestic computing power platforms, including Huawei Ascend, PingTouGe, and Moore Threads. The acquisition of Zhongke Jiahe will not only directly improve Zhipu's unit Token inference cost and output quality but also provide core underlying compiler technology support for its previously rumored self-developed custom AI inference chip plan.

According to "Intelligent Emergence," multiple independent sources reported that by July 2026, China's leading large model company, Zhipu, had achieved an annual recurring revenue (ARR) of $1 billion. Insiders revealed that between January and July 2026, Zhipu's ARR surged 15 times in just six months. The growth from $100 million ARR to $1 billion took Zhipu only 5 months, surpassing the 15 months it took the American AI lab Anthropic to reach a similar stage. As of the time of publication, Zhipu has not responded to the aforementioned financial data.The rapid rise in Zhipu's revenue is primarily attributed to its concentrated investment in coding and reasoning capabilities. Financial reports show that in the first quarter of this year, even though the API call price for GLM increased by approximately 83%, its overseas subscription prices approached those of Anthropic's Claude Code, yet its total call volume still grew by about 400% against the trend. In June of this year, Zhipu launched its latest open-source large model, GLM-5.2, which has matched or even surpassed mainstream cutting-edge models on several core metrics.Currently, AI coding and video generation have become the fastest commercialized and most revenue-generating tracks for large models globally. As the demand in the coding market becomes more certain, industry competition is intensifying. Domestically, MiniMax released the M3, which focuses on enhancing coding capabilities, in June, while the Dark Side of the Moon launched the K3 open-source model with 2.8 trillion parameters on July 16; internationally, with OpenAI merging ChatGPT and CodeX, it has also shown a momentum to catch up with Anthropic in the coding field, and the global battle for AI productivity tools continues to escalate.