Alibaba Begins Selling Quark AI Glasses in China, Its First Qwen-Powered Wearables(Yicai) Nov. 28 -- Alibaba Group Holdings has launched its Quark AI Glasses, the Chinese internet giant's first product to integrate its Qwen large language model.
Alibaba started selling six models yesterday, including the S1 and the G1, which start at CNY3,799 and CNY1,899 (USD535 and USD270), respectively. The G1, priced lower because it lacks a display, weighs only 40 grams but has the same chip, acoustics, and camera as the S1.
The S1 is available on major Chinese e-commerce platforms and in hundreds of brick-and-mortar partner stores across 82 cities. The G1 is open for pre-orders. They will compete with Ray-Ban Meta’s smart glasses.
The Quark AI Glasses are the first hardware product powered by Qwen, Alibaba’s flagship artificial intelligence model, marking its first step from digital screens into the physical world, Vice President Wu Jia said at the product launch event.
With their launch, the Hangzhou-based company has begun enhancing Qwen’s use in the consumer market through the synergy of software, hardware, and ecosystem integration.
Alibaba is committing heavily to AI technologies, having unveiled a three-year CNY380 billion (USD53.7 billion) investment plan in February. Chief Executive Eddie Wu said earlier this week that that plan now seems a bit conservative.
“Unlike other hardware, the competition among developers of AI glasses lies in their systematic capabilities,” said Song Gang, Alibaba’s head of smart devices. Smart glasses require end-to-end capabilities, and the Quark AI Glasses can fully leverage Quark's vast search data, the Qwen LLM, and integrate with the firm's proprietary ecosystem, Song pointed out.
In addition to integrating Alibaba's apps, including Alipay, Amap, Taobao, and Fliggy, the glasses will also feature external resources such as QQ Music, NetEase Cloud Music, and Flight Manager in the future, product lead Jin Xian noted, adding that the aim is to also build a developer ecosystem by supporting the model context protocol.
Editors: Tang Shihua, Martin Kadiev