Zhipu’s September 2 Tmall launch let shoppers buy GLM coding capacity as a prepaid subscription: ¥118/month buys Lite with 10,000 weekly credits, while Max costs ¥1,078 for 140,000. The store offers Lite, Pro, Max and a ¥598 per seat team plan, sold on monthly, quarterly and annual cycles; all use a five hour plus w...
Published byEdited with GPT-5.6 TerraImages generated with GPT Image 2
Research answer

Create a landscape editorial hero image for this Studio Global article: How did Zhipu AI’s September 2, 2026 launch of a Tmall flagship store for GLM Coding Plan token subscriptions—its first placement of a core. Article summary: Zhipu turned AI-model usage into a retail, prepaid utility: a Taobao shopper could select a tier, pay through a familiar e-commerce checkout, and receive a defined allowance of coding-model credits—much like topping up a. Topic tags: general, general web, user generated, documentation. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks,
Buying access to an advanced AI model has traditionally meant opening a developer account, choosing an API pricing scheme and monitoring usage in a technical dashboard. Zhipu AI’s Tmall flagship store takes a more consumer-retail approach: shoppers can select a GLM Coding Plan, check out through Taobao and receive a defined allowance of model credits. In practical terms, it resembles renewing a phone-data package or a video subscription more than negotiating an enterprise software contract. 1
The store launched on September 2, 2026, offering individual Lite, Pro and Max subscriptions, plus a standard team seat. Listings reported monthly, quarterly and annual purchase options, rather than a Tmall-only price cut. 1
6
| Plan | Monthly listed price | Five-hour credits | Weekly credits |
|---|---|---|---|
| Lite | ¥118 | 2,000 | 10,000 |
| Pro | ¥538 | 12,000 | 60,000 |
| Max | ¥1,078 | 28,000 | 140,000 |
| Team standard seat | ¥598 per seat | Not specified in the provided store reporting | Not specified in the provided store reporting |
The three personal plans use Zhipu’s GLM-5.3 model and support more than 20 coding-agent tools, including ZCode, Claude Code and Codex, according to reporting on the store. 1
7
That is the consumer-product shift: the customer is not buying a model deployment or an open-ended service relationship. They are buying an identifiable package of capacity, at a familiar price point and through a familiar checkout flow.
Zhipu’s documentation defines two limits for each personal plan:
This design separates short bursts of intensive coding work from a subscriber’s total weekly use. A Lite subscriber, for example, has 2,000 credits available in the rolling five-hour window but cannot exceed 10,000 credits in the week. Pro and Max raise both ceilings. 17
Credits are connected to underlying usage rather than representing a fixed number of prompts. Zhipu’s documentation describes a calculation based on input tokens, cached input tokens, output tokens and applicable multipliers. 17 Prompt counts therefore vary with the task, context length and model behavior. As a rough illustration, Zhipu’s July plan-update notice estimated up to about 80 prompts per five hours and 400 per week for Lite; Pro was estimated at about 400 and 2,000, while Max was estimated at about 1,600 and 8,000.
18
The important benefit is visibility. Instead of a subscription that appears unlimited but is governed by unclear throttling, the plan shows both near-term capacity and a weekly ceiling. That does not eliminate limits; it makes the limits easier to understand and budget for.
A report on the product’s history says the Coding Plan debuted in September 2025, faced quota rationing from January 2026 after demand increased around GLM-4.7, and returned to open sale on July 31 following the move to the credit system. Those earlier milestones are reported context rather than independently confirmed by the official documentation provided here.
What is directly documented is the late-July usage revision and its two-window quota structure. 18 The subsequent Tmall listing matters because it moves the plan from a specialist AI purchase into an ordinary e-commerce catalog. A coding subscription can now be discovered, compared and bought in the same environment consumers already use for everyday goods.
The storefront aligns with a sharp increase in Zhipu’s usage-based revenue. For the six months ended June 30, 2026, Zhipu reported revenue of ¥953.9 million, up 399.7% year over year. Open-platform and API revenue was ¥825.2 million—86.5% of total revenue, compared with 15.2% in the prior-year period.
Those figures do not prove that Tmall sales will be a major revenue driver. But they show why a standardized credit product is strategically coherent: the company’s reported business mix has shifted toward access delivered and billed through cloud platforms and APIs. A packaged coding plan is another way to sell repeatable consumption of that same kind of capacity.
Zhipu’s store did not remain an isolated experiment. Tmall subsequently introduced an “AI Space Station” carrying services from Alibaba Cloud, Zhipu, Kimi and MiniMax. The section offers token plans, coding subscriptions and pay-as-you-go packages, with delivery through redemption codes or direct account top-ups.
The retail framing coincides with rapid growth in token use. China’s average daily token calls exceeded 140 trillion in March 2026, according to figures cited by Global Times. Major telecom operators had also begun selling consumer token packages, including plans starting at ¥9.9 a month, according to People’s Daily Online.
That does not mean AI tokens have become as simple or interchangeable as mobile data. Model quality, context size, output length, tool integrations and credit multipliers still affect what a customer receives. But the commercial interface is changing: AI inference is increasingly packaged as a recurring, measurable retail utility.
For a developer choosing a GLM Coding Plan, the decision is less about a vague promise of “AI access” and more about matching a workload to two explicit ceilings:
The deeper change is not the storefront alone. Zhipu is presenting coding-model capacity in units that consumers can recognize, compare and replenish—bringing AI access closer to the retail logic of a top-up plan.
Studio Global AI
This page includes a source-backed answer you can continue inside Studio Global.
Zhipu’s September 2 Tmall launch let shoppers buy GLM coding capacity as a prepaid subscription: ¥118/month buys Lite with 10,000 weekly credits, while Max costs ¥1,078 for 140,000.
Zhipu’s September 2 Tmall launch let shoppers buy GLM coding capacity as a prepaid subscription: ¥118/month buys Lite with 10,000 weekly credits, while Max costs ¥1,078 for 140,000. The store offers Lite, Pro, Max and a ¥598 per seat team plan, sold on monthly, quarterly and annual cycles; all use a five hour plus weekly credit system.
The launch fits a wider Chinese retail shift: Tmall’s AI section now sells token plans alongside other consumer technology, while China reported more than 140 trillion daily token calls in March 2026.