How to buy Zhipu GLM Coding Plan: personal tiers run Lite ¥118, Pro ¥538 and Max ¥1078/mo with 20% off quarterly and 30% off annual, and team plans start at ¥598/seat/mo. Most individual developers should pick Pro — 12,000 credits/5h and 60,000/week is about 6x Lite and covers daily high-frequency development; light trials fit Lite, and Max only makes sense for multi-project or heavy agent workloads. This guide walks through the purchase path, every tier's price and its credit quota, with a selection verdict at the end.
Where to Buy: Official Entry and Purchase Flow
There is only one entry: the GLM Coding product page (bigmodel.cn/glm-coding), where both personal and team plans start (source). The flow is three steps: pick a tier, subscribe and pay, then grab the Coding-specific API key from the console. Personal plans support continuous monthly, quarterly and annual billing; team plans bill per seat monthly or annually.
This key is a separate system from the Zhipu API Open Platform pay-as-you-go key: the Coding key works only inside officially supported coding tools (Claude Code, Cursor, Cline, TRAE and 20+ more), configured with a Coding-specific base URL (Claude Code uses https://open.bigmodel.cn/api/anthropic). For self-hosted backends, use the Open Platform API key — don't point a Coding key at them.
Pricing and Quota: All Tiers Compared
The price comparison verdict: 20% off quarterly, 30% off annual — annual is cheapest, with Lite working out to ¥82.6/mo (source). Personal list prices:
| Tier | Monthly | Quarterly (20% off) | Annual (30% off) |
|---|---|---|---|
| Lite | ¥118 | ¥94.4/mo | ¥82.6/mo |
| Pro | ¥538 | ¥430.4/mo | ¥376.6/mo |
| Max | ¥1078 | ¥862.4/mo | ¥754.6/mo |
Annual is paid up front (Lite ¥991.2/yr, Pro ¥4,519.2/yr, Max ¥9,055.2/yr). Month-to-month makes sense for a short trial; if you're staying long-term, go annual and save 30% — the most concrete deal available.
Quota does not bill by token — it uses a credit system with quota issued on 5-hour and weekly cycles. Personal credit quota:
| Tier | 5-hour credits | Weekly credits | vs Lite |
|---|---|---|---|
| Lite | 2,000 | 10,000 | Baseline |
| Pro | 12,000 | 60,000 | 6x |
| Max | 28,000 | 140,000 | 14x |
Once quota is exhausted it resets on the next cycle without auto-charging your balance. Off-peak hours (outside Mon–Fri 14:00–18:00 UTC+8) burn model calls at 50% of base credits — evenings and weekends stretch your quota much further.
Credit Coefficients: What GLM-5.3 / 5-Turbo / 4.7 Actually Burn
Within the same plan, the model you pick decides your burn rate — GLM-5.3 is the most expensive, with output 50% pricier than GLM-4.7. The three available models:
| Model | Input | Cached | Output | Notes |
|---|---|---|---|---|
| GLM-5.3 | 6.9 | 1.7 | 24 | Flagship, first choice for complex engineering |
| GLM-5-Turbo | 5.7 | 1.5 | 21 | Middle ground for long chains |
| GLM-4.7 | 4.6 | 1.2 | 16 | Daily workhorse, cheapest |
For the same 1M output tokens, GLM-5.3 burns 24 credits while GLM-4.7 burns 16. The practical combo: GLM-4.7 for daily Q&A and small-repo iteration, GLM-5.3 for complex agentic coding and terminal work. Web search, page reading and open-repo MCP each cost a flat 1.2 credits, sharing the same pool as model calls.
Team vs Personal: Where They Diverge
Team plans meter independent credits per seat and target teams that need multi-user management and unified billing; personal plans are one person, one quota. The two team tiers:
| Tier | Monthly | Annual (10% off) | Credits per seat (5h/week) |
|---|---|---|---|
| Standard | ¥598/seat | ¥538.2/seat | 15,000 / 66,000 |
| Advanced | ¥1198/seat | ¥1078.2/seat | 35,000 / 155,000 |
Two seats minimum. Team plans include seat and permission management, usage dashboards, consolidated billing and invoicing, and data not used for training by default; exhausted quota can enable overage billing at 90% of API list price (limited promo, admin-enabled). The purchasing admin does not occupy a seat by default.
How to Choose: Decision by Scenario
Three direct verdicts — match your workload and go.
- Light trial, small repos, occasional Q&A → Lite (¥118/mo). 2,000/5h handles light iteration; use it to validate whether the GLM + MCP workflow fits you.
- Daily high-frequency individual development with a stable search-read-edit loop → Pro (¥538/mo). The officially highlighted tier: 12,000/5h and 60,000/week is about 6x Lite, with official guidance supporting 1–2 parallel project-level agent workflows.
- Multi-project, large repos, heavy agent workloads → Max (¥1078/mo). 28,000/5h and 140,000/week is about 14x Lite, with peak-time resource priority.
For teams: Standard covers daily work on medium repos; Advanced fits high-concurrency, multi-project work that needs peak-time guarantees. And run the monthly-vs-annual math before buying — annual saves 30% for long-term users.
Caveats
- Plans work only inside officially supported tools; seat sharing or unsupported tools may trigger risk controls.
- GLM-5.3 weights aren't open-sourced yet (Z.ai targets late August), so third-party tooling may shift.
- Legacy GLM-5.2/5.1 calls auto-route to GLM-5.3 — if you budgeted by GLM-4.7 coefficients, the same tokens now burn more credits.