Guides · dev · glm coding plan lite

GLM Coding Plan: Lite, Pro, and Max, in credits not slogans

Published 16 Oct 2026. Updated 16 Oct 2026. Researched 16 Oct 2026.

GLM Coding Plan is a subscription that meters credits, not a discount coupon on the token API. The overview says it starts at 18 USD a month and that every plan has both a 5-hour limit and a weekly limit.

Plan 5-hour credits Weekly credits
Lite 2,000 10,000
Pro 12,000 60,000
Max 28,000 140,000

Those credit counts are from the overview, accessed 16 Oct 2026. The dollar figures below are from the legacy-plan migration notice, which is dated as a publication of 21 Apr 2026 and still describes them as the current standard price:

Plan Standard price on that notice
Lite $18
Pro $72
Max $160

The same notice shows a separate “discounted price” table with higher numbers ($48.60 / $194.40 / $432). Read that as a migration worked example, not as a second public price list, and then check the subscribe page yourself. We did not complete a purchase. A secondary tracker we saw while researching listed $18 / $80 / $168. When two pages disagree, the vendor’s checkout wins.

Credits are not tokens

Z.AI does not, on the overview page, publish how many tokens one credit buys. Anyone who converts “2,000 credits” into “a week of Claude Code” is guessing. The only honest statement is: Lite refills 2,000 credits per 5-hour window and 10,000 per week, and both caps apply.

The model docs say off-peak calls, including weekends, can consume fewer points on the coding plan. That is a points rule, not the API’s token price. Do not mix it up with DeepSeek’s off-peak token discount. They are different products.

The API is still there if the plan is a bad fit

Pay-as-you-go GLM-5.3 is $1.40 / $4.40 per million on Z.AI’s pricing page, with cached input at $0.26. Flash is $0.15 / $0.50. A light user can spend less on the API than on Lite. A heavy user hits the weekly credit cap and then has to wait or pay tokens anyway.

Illustrative Coding-agent session costGLM-5.3$1.39GLM-5.3-Flash$0.156Anthropic: Claude Opus 5.5$2.94OpenAI: GPT-6.1 Sol$1.47
Coding-agent session: 3 million input tokens, 90% served from cache, 60 thousand output tokens. Cache-write fees are not included. Chinese rows use the lab published rate (DeepSeek uses the peak rate). Comparators are OpenRouter list prices for Anthropic: Claude Opus 5.5 ($8.00/M blended) and OpenAI: GPT-6.1 Sol. Snapshot 2026-10-16. Cache-write fees are excluded. A missing cache price is billed as a full input token.
ModelSessionvs Opusvs SolCache price
GLM-5.3$1.3953% cheaper6% cheaperpublished
GLM-5.3-Flash$0.15695% cheaper89% cheaperpublished
Anthropic: Claude Opus 5.5$2.94——published
OpenAI: GPT-6.1 Sol$1.47——published

The percentages in this table compare session cost. The price table's percentages compare blended token price. They answer different questions.

The session in that chart is illustrative: 3 million input tokens, 90 percent cache hits, 60 thousand output tokens, no cache-write fee. It is not your repo. It is a way to see that Flash and the flagship are not the same bill, and that the flagship’s official rate is the one to compare with a $18 plan, not the cheap OpenRouter list price.

Wiring it to a coding agent

Z.AI documents an Anthropic-compatible base URL:

ANTHROPIC_BASE_URL=https://api.z.ai/api/anthropic

Put the key in the header the docs specify. We did not run an agent against it. The Claude Code guide collects the base URLs we were willing to copy.

Who it is for

Lite is a capped experiment, not a Claude Max replacement. Pro and Max raise the cap sixfold and fourteenfold against Lite’s 5-hour number, and the price on the migration notice rises faster than that from Lite to Pro. If your week is bursty, the weekly cap matters more than the 5-hour one. If your week is steady and small, the token API is easier to explain on an invoice.

Related: coding plans compared, Z.AI API, session cost.

Questions

How much is GLM Coding Plan Lite?

Z.AI's migration notice lists a current standard price of $18 for Lite, $72 for Pro, and $160 for Max. A third-party tracker disagrees on Pro and Max. Confirm the number on the subscribe page before you pay.

What happens when the credits run out?

The overview sets two caps at once: a 5-hour credit window and a weekly credit window. Hitting either one stops you, even if the other still has credits.

Sources

  1. GLM Coding Plan overview, accessed 2026-10-16
  2. GLM Coding Plan legacy migration notice, accessed 2026-10-16
  3. Z.AI pricing, accessed 2026-10-16
  4. GLM-5.3 docs, Anthropic base URL, accessed 2026-10-16