Z.ai gives GLM Coding Plan subscribers a quota refill after one year

Z.ai announced on September 1 that every current GLM Coding Plan subscriber will receive a Reset Card that refills both the rolling five-hour allowance and the weekly quota, with redemption requiring a signed-in ZCode account. The promotion marks the one-year anniversary of the plan's launch and aims to retain developers while driving adoption of ZCode, Z.ai's development environment. Co-founder Zhang Peng has emphasized coding as the company's breakthrough point, linking GLM-5 iteration to growth in annual recurring revenue.

Z.ai gives GLM Coding Plan subscribers a quota refill after one year Every current subscriber gets a Reset Card for weekly and five-hour limits, with redemption tied to a signed-in ZCode account. By RuntimeWire Staff /author/runtimewire-staff · Published Primary source: Z.ai https://x.com/Zai org/status/2094769612730532172 Why it matters Z.ai is using quota relief to retain paying developers and pull them into ZCode, turning a subscriber perk into distribution for its own coding agent. Z.ai https://z.ai/?ref=runtimewire co-founder Zhang Peng https://www1.hkexnews.hk/listedco/listconews/sehk/2025/1230/2025123000017.pdf?ref=runtimewire is marking the GLM Coding Plan https://z.ai/subscribe?invitedialog=true&ref=runtimewire 's first anniversary with a practical gift: one more turn at the meter. Z.ai said on September 1 that every current GLM Coding Plan subscriber will receive a Reset Card. The card refills both the rolling five-hour allowance and the weekly quota, letting developers resume coding-agent sessions without waiting for the normal usage windows to refresh. The promotion comes one year after the GLM Coding Plan's reported September 1, 2025 launch. It also lands during a sustained effort by Zhang and his co-founders to turn coding from a showcase for the GLM model family into a recurring subscription business. A useful gift with a ZCode requirement The Reset Card addresses an ordinary constraint of AI coding subscriptions: long agent runs can consume an allowance before the work is finished. Z.ai's current plans https://docs.z.ai/devpack/overview?ref=runtimewire track a five-hour prompt pool alongside a weekly limit, and supported tools share the subscription's available quota. Z.ai's announcement says the card replenishes both the rolling five-hour quota and the weekly quota. The supplied documentation indicates that users must be signed in to ZCode and link a Coding Plan account to receive and use reset cards. A subscriber working solely through an API key will not receive a card until the account is linked inside ZCode. ZCode's documentation https://zcode.z.ai/en/docs/usage-stats?ref=runtimewire says a redeemed card returns the selected quota to 100%. The reset panel displays each card's expiration and uses the earliest-issued card first when an account holds several. That design gives subscribers immediate additional usage while drawing them toward ZCode, the development environment Z.ai has been building around its own models. The GLM Coding Plan can connect to more than 20 supported coding tools https://zcode.z.ai/en/docs/configuration?ref=runtimewire , allowing developers to use GLM through products they already know. Reset cards, however, are distributed and managed inside ZCode. Z.ai gets retention and product distribution from the same giveaway. Subscribers get relief at the exact moment the service has become most valuable to them: when an agent has consumed its quota in the middle of a task. Zhang Peng made coding the commercial wedge The promotion follows a strategy Zhang laid out for shareholders earlier this year. In an April founder's statement https://www1.hkexnews.hk/listedco/listconews/sehk/2026/0419/2026041900085.pdf?ref=runtimewire , he wrote that Z.ai had identified "coding capabilities as our breakthrough point" and connected rapid GLM-5 /models/z-ai/glm-5 iteration with growth in annual recurring revenue. Z.ai did not attach a figure to that ARR claim in the statement. Zhang's focus on coding grew out of a research organization rather than an editor business. He worked at Tsinghua University from 2005 through 2020, and Z.ai was founded in 2019 as a commercialization of technology developed at the university. Z.ai's founding group also includes Tsinghua researchers Jie Tang and Juanzi Li, whose Knowledge Engineering Group worked on knowledge graphs, semantic mining and language understanding before large models became a global commercial category. That background helps explain Z.ai's route into developer tools. Zhang has framed the next stage of the model market around AI systems that act for longer periods, retain context and complete multi-step work. Coding offers a measurable proving ground for that thesis: repositories expose whether an agent can plan, edit, run tools, find its mistakes and continue after a failure. A quota reset is a small product decision inside that larger bet. Longer-running agents require users to trust both the model and the billing system. A developer who repeatedly reaches a limit has demonstrated demand, though the same event can become a reason to switch providers. Reset cards turn that point of friction into another session. A year of increasingly aggressive distribution Z.ai spent August tightening the link between its models, subscription and first-party coding environment. On August 11, the company said ZCode had reached 1 million users https://x.com/Zai org/status/2087040814677725262?ref=runtimewire and reset Coding Plan limits. The self-reported figure did not distinguish registered, active or paying users. Ten days later, Z.ai offered 100 million GLM-5.3 /models/z-ai/glm-5.3 tokens to each of 50,000 eligible new ZCode accounts, an advertised maximum of 5 trillion tokens. The terms were documented in contemporaneous reporting on the campaign https://groundtruth.day/news/glm-5-3-shipped-with-a-ledger-of-2436-security-findings.html?ref=runtimewire . Z.ai then released GLM-5.3's open weights https://z.ai/blog/glm-5.3?ref=runtimewire on August 28 after delaying them for safety evaluation and hardening because of unexpectedly strong cyber capabilities, according to Axios reporting on the delay https://www.axios.com/2026/08/14/china-open-source-ai-glm-53?ref=runtimewire . The anniversary card continues the same distribution play with a different audience. The token campaign targeted new users. Tuesday's reset rewards people already paying for the Coding Plan and gives them a reason to open ZCode again. Z.ai is competing in a coding-agent market where developers can move among models, command-line agents and editors with relatively little ceremony. The GLM Coding Plan's appeal rests partly on that portability: subscribers can use Z.ai's models across supported tools instead of committing to one proprietary interface. Z.ai still needs its own interface to hold the customer relationship, present new features and make model upgrades visible. Reset cards serve that purpose without closing the plan off from outside tools. Zhang's approach is to sell access broadly, then make ZCode the place where the subscription feels most generous. One year in, Z.ai is celebrating by giving paying developers additional compute and asking them to collect it on Z.ai's home turf.