Z.ai stock surged after launching an AI model running entirely on Chinese chips

Z.ai stock surged after launching an AI model running entirely on Chinese chips

Published 19 days ago

Free daily briefing on global business news.

The company says it handled all online traffic for the model on 100,000 domestically made chips, a claim CNBC could not independently verify

Bloomberg / Getty Images

Z.ai released a new AI model Wednesday built to run on Chinese chips, sending its Hong Kong-listed stock up more than 12% on Thursday.

The model, called GLM-5.3-Flash, is a lower-cost version of Z.ai's flagship offering. The company said it used 100,000 Chinese-made chips to handle all online requests, including during a one-week preview period when the model circulated under the code name "Ox Alpha." Neither claim could be confirmed independently, and Z.ai would not say which chipmakers supplied the hardware, according to CNBC.

GLM-5.3-Flash has 320 billion total parameters and 18 billion active parameters, the company said. The model places 10th on the Artificial Analysis Intelligence Index, finishing above DeepSeek V4 Pro Max. Z.ai priced the model at $0.045 per task, which it said is roughly one-tenth the cost of comparable offerings.

While still operating under its Ox Alpha preview name, the model racked up over 11 trillion tokens within its first three days on OpenRouter — a record opening for the platform, according to the South China Morning Post. OpenRouter data from Thursday showed it had climbed to the top position among coding models on the platform, with its weekly token count representing close to 31% of all traffic.

Counterpoint Senior Research Analyst Ivan Lam said Z.ai's cluster probably includes Huawei Ascend processors as well as chips from other vendors, and he pointed to a broader trend of Chinese firms tightening ties across the hardware and software stack. "Chinese AI model developers have continued to allocate more resources and investment toward AI servers and computing infrastructure built on domestic chips," Lam said.

Z.ai said it built a dedicated inference engine on top of the SGLang framework and achieved a 3x improvement in serving performance over its initial baseline on the same hardware, reaching cost and efficiency levels it said are comparable to mainstream Nvidia $NVDA GPUs.

The launch comes as China pushes to reduce dependence on U.S. semiconductors. Export controls have sharply curtailed Nvidia's ability to ship advanced chips into China, prompting Huawei and a growing number of local semiconductor firms to accelerate their own competing offerings. Nvidia CEO Jensen Huang has said American companies should be free to use Chinese open-source AI models, arguing that cheaper and more accessible models ultimately drive demand for chips and data centers.

Z.ai is scheduled to report results for the first half of the year on Monday. The company's stock has climbed more than 800% since its Hong Kong listing in January.

Join 500,000+ readers who start their day with Quartz.

By subscribing, you agree to our Terms of Service and Privacy Policy.

Global business news for a smarter world

© 2026 Quartz Media, Inc. All rights reserved.