All coding subscriptions

GLM Coding Lite

Zhipuentry
$18
Last checked: 2026-10-19

This tier is sold in China too, at a price the vendor sets for that market: GLM Coding Lite(国内)¥118 per month

First limit you hit
No published credit rate
Equivalent API value
—
Value multiple
—
When it runs out
Stops until the window resets
Scenario
Interactive use only

Allowances

2,000credits per 5 hours
›The vendor's own wording
The "Usage Credit Allowance" table gives Lite "2,000" under "5-Hour Credits". "5-hour credits: Dynamically refreshed; credit quota resets 5 hours after consumption."
Open the source page
10,000credits per 7 days
›The vendor's own wording
Lite "10,000" under "Weekly Credits". "Weekly credits: Activated upon subscription; resets every 7 days."
Open the source page
48,000,000–97,000,000tokens per week (vendor estimate)
›The vendor's own wording
The "Estimated Token Allowance" table gives Lite "48–97" M tokens/week for GLM-5.3 at a 95% cache hit rate, the lowest of its six rows. "Minimum token allowance: All usage occurs during peak hours"; "Maximum token allowance: All usage occurs during off-peak hours".
Open the source page

Models included

ModelCapabilityBurn rate
GLM-5.3defaultuntiered1Multipliers 6.9 in / 1.7 cached / 24 out. "During off-peak hours, model usage is charged at 50% of the standard credit rate." Peak: Mon–Fri 14:00–18:00 UTC+8.
GLM-5.3-Flashuntiered0.333Multipliers 2.3 in / 0.56 cached / 8 out, with the same off-peak halving. Includes the vision-understanding MCP.

Subscription versus your own API key

This compares two ways of working the same way — in an IDE or CLI, where BYOK tools make a direct API key a real option. It is not a claim that subscriptions are cheaper than the API in general.

ModelEffective price / 1M tokensvs API rateBreakeven
GLM-5.3untiered——18.5M
GLM-5.3-Flashuntiered——179.8M

Usage terms

Quoted from the vendor's own terms. We do not give legal advice.

The plan is confined to Zhipu's list of supported coding tools: "GLM Coding Plan 仅限在官方支持的指定工具与产品环境中使用。在除规定工具外调用 API,不可享用 Coding 套餐的额度。" Integrations go to the standard API instead: "如需在自建应用、网站、机器人、SaaS 产品等场景中通过 API 集成模型能力,请使用 智谱提供的标准 API 服务,并根据对应协议计费。" OpenClaw is allowed but demoted — "套餐支持 OpenClaw 使用,但采用次级调度与尽力交付策略,Coding Agent 任务享有 资源抢占优先权".

Change history

  1. 2026-10-14Recorded from the seed file
  2. 2026-07-28Recorded from the seed file