Zhipu opens official Tmall flagship store to sell GLM Coding Plan subscriptions

Chinese AI lab Zhipu has taken its developer products to the consumer marketplace. On September 4, the company formally opened an official flagship store on Alibaba's Tmall, currently selling GLM Coding Plan subscriptions — personal Lite, Pro and Max tiers plus a standard team seat — with monthly, quarterly and yearly billing. Coming days after Zhipu disclosed its half-year results, the move is widely read as a new push in commercialization.

The prices are not Tmall exclusives: Lite costs 118 yuan a month, Pro 538 yuan, Max 1,078 yuan, and the team standard seat 598 yuan, identical to Zhipu's official website. The plans run on GLM-5.3 and can connect to more than 20 agentic coding tools including ZCode, Claude Code and Codex.

Developers have typically bought API access and tokens through a model vendor's own site or open platform; Zhipu has now turned them into off-the-shelf goods on a Taobao marketplace shelf. Early sales figures for the store look modest, but the move has already generated attention — an AI lab selling tokens on Tmall makes for a catchy headline.

More significant than the store's initial orders is what the channel shift says about Zhipu's business model. Days earlier, Zhipu reported first-half 2026 revenue of 954 million yuan, up nearly 400 percent year over year, with open-platform and API revenue reaching 825 million yuan, or 86.5 percent of the total, compared with just 15.2 percent in the same period last year.

The figures show Zhipu increasingly earns its money from usage volume. With model capabilities built, the practical questions become how to make it easier for developers and small teams to pay, renew and invoice, and how to run promotions and acquisition campaigns around shopping festivals like Double 11 and 618. Tmall offers a mature consumer infrastructure for exactly that.

It is also a shift in how AI services are sold. Discussion of large-model commercialization used to center on API pricing, enterprise procurement and private deployment. With Coding Plan and agent bundles maturing, large-model capabilities are being repackaged as standardized, subscribable, promotable digital goods.

In the near term, how many subscriptions Zhipu's store sells matters less than whether other vendors follow it onto e-commerce platforms, complete with festival discounts, membership perks and enterprise coupons. The underlying signal is that model makers are learning to turn technology into products — and then to sell those products.