Z.ai GLM

z.ai · Coding subscriptions, Compute · 2 records · JSON

Z.ai GLM Flash Computecaveatemailno cardDrequest-rate
Z.ai GLM Coding Plan Coding subscriptionspaidcardcardDrequest-rate

Z.ai GLM Flash Compute

caveatemailno cardDrequest-rate

docs.z.ai/guides/overview/pricing · inference API · compute/z-ai-glm-flash · JSON · compare

Free tier	Two models are priced Free on every column of the official pricing table, with no credit and no trial window: GLM-4.7-Flash and GLM-4.5-Flash, both 'Free \| Free \| Free \| Free' for input, cached input, cached input storage and output. The paid GLM-5.3-Flash sits beside them at USD 0.15 input and USD 0.50 output per 1M tokens, which is the rate to price the free lane against. unit none; metered on wall-clock; recurring-free-model
Price	none stated
Per-object cap	not applicable
Retention	not applicable
Egress	not stated
Identity	email not published
Limits	request-rate no quota is published at all: the FAQ points rate limits at z.ai/manage-apikey/rate-limits, which is behind a login, so the concurrency ceiling is unknown until an account exists and it is the first thing to measure. source no quota is published at all: the FAQ points rate limits at z.ai/manage-apikey/rate-limits, which is behind a login, so the concurrency ceiling is unknown until an account exists and it is the first thing to measure.
Binding	request-rate no quota is published at all: the FAQ points rate limits at z.ai/manage-apikey/rate-limits, which is behind a login, so the concurrency ceiling is unknown until an account exists and it is the first thing to measure.
Interface	rest inferred from the tooling the provider documents; not probed
Tooling	not stated
Composes with	a free model with an OpenAI-compatible endpoint is the cheap first pass in a two-stage pipeline: GLM-4.7-Flash classifies or filters, and only the survivors go to a metered model.
Reliability	no reliability event recorded
Notes	The row that would settle this is the rate limit, and it is not public: three routes tried, the pricing doc (no limits), the FAQ (points at a login-gated page), and docs.z.ai/api-reference/rate-limit (redirects to z.ai/manage-apikey/rate-limits, which returned a marketing shell with no table). The docs are served in English and the API takes an international card or no card at all, so this is one of the few Chinese lanes with no national-ID step. Not measured: no account.
Evidence rung	documented
Checked	2026-10-22 (live); stale on 2026-11-21

Evidence (3)

date	source	what it proved
2026-10-22	docs.z.ai/guides/overview/pricing	'\| GLM-4.7-Flash \| Free \| Free \| Free \| Free \|' and '\| GLM-4.5-Flash \| Free \| Free \| Free \| Free \|' in the text models table, quoted
2026-10-22	docs.z.ai/guides/overview/pricing	GLM-5.3-Flash priced USD 0.15 input and USD 0.50 output per 1M tokens, which is the paid comparator for the free Flash lane, quoted
2026-10-22	docs.z.ai/help/faq	'How can I check the rate limits? You can check the rate limits from here: https://z.ai/manage-apikey/rate-limits', quoted; the page requires a login

More fields (1)

{
 "wave2": {
  "class": "perpetual-free-allowance",
  "free_tier_kind": "recurring-free-model",
  "shape": {
   "vcpu": null,
   "ram_gib": null,
   "gpu_class": null,
   "gpu_count": null
  }
 }
}

Unknown or withheld values (9)

$.egress.metered egress-unknown
$.egress.terms not-stated
$.identity_cost.detail process
$.per_object_cap.notes process
$.per_object_cap.stated not-applicable
$.reliability_note none-observed
$.retention.notes process
$.retention.stated not-applicable
$.tooling not-stated

Z.ai GLM Coding Plan Coding subscriptions

paidcardcardDrequest-rate

docs.z.ai/devpack/overview · coding subscription · coding-subs/zai-glm-coding-plan · JSON · compare

Free tier	no free tier; Lite from USD 18 a month unit none; metered on none; usage is metered in credits, and the credit formula is published
Price	none stated
Per-object cap	not applicable: a subscription meters usage, not objects
Retention	not applicable
Egress	not applicable: the vendor serves the inference
Identity	card a paid subscription
Limits	request-rate Lite 2,000 five-hour credits and 10,000 weekly; Pro 12,000 and 60,000; Max 28,000 and 140,000 source request-rate credit cost = (input tokens x input multiplier + cached input x cached multiplier + output x output multiplier) / 10,000; GLM-5.3 multipliers are 6.9, 1.7 and 24 source
Binding	request-rate
Interface	rest, cli supported in Claude Code, Codex, OpenCode, Cline, ZCode and others
Tooling	any Claude-Code-compatible client, plus the named supported tools
Composes with	the credit formula lets a token budget be converted to a plan, which almost no other plan in this topic allows
Reliability	no reliability event recorded
Notes	Neither pair is verifiable today: the subscribe page renders the tier names and not the prices. Both figures are therefore unverified and the record carries no price for those tiers.
Evidence rung	documented the credit allowances and the per-model multipliers are quoted from the vendor documents today. Lite is USD 18; the Pro and Max prices were not readable. verified_at is the instant its evidence URLs were fetched on 2026-10-22, read from the fetch log, not typed.
Checked	2026-10-23 (live); stale on 2026-11-22

Evidence (2)

date	source	what it proved
2026-10-22	docs.z.ai/devpack/overview	Lite from USD 18 a month; 5-hour and weekly credit allowances of 2,000 and 10,000 on Lite, 12,000 and 60,000 on Pro, 28,000 and 140,000 on Max; credit formula with per-model multipliers; 5-hour credits reset five hours after consumption and weekly credits every seven days
2026-10-22	z.ai/subscribe	Lite, Pro and Max tiers exist and Pro carries six times Lite usage; the price values did not render in any extraction route

Unknown or withheld values (3)

$.notes partly-held
$.reliability_note none-observed
$.retention.stated not-applicable

Corrections

2026-10-22 Kimi and zai ladders do not match