What Is the GLM Coding Plan in 2026?
The GLM Coding Plan is z.ai's subscription for using its GLM models inside AI coding and agent tools. In October 2026 it runs on GLM-5.3 and GLM-5.3-Flash, comes in Lite, Pro and Max for individuals plus a Team plan, and gives you a credit allowance that refreshes every 5 hours and every week, instead of charging per token.
This guide was first written for GLM-5.2. z.ai released GLM-5.3 on 18 August 2026, and its docs say Coding Plan requests for GLM-5.2 or GLM-5.1 are now routed to GLM-5.3 automatically. Prices and allowances below are the October 2026 figures.
z.ai says the plan works with more than 20 agent tools, including its own ZCode IDE, Claude Code, Codex, OpenCode and Cline, and covers office tasks such as research and data analysis as well as coding.
How Much Does the GLM Coding Plan Cost?
The GLM Coding Plan costs $18 a month for Lite, $80 for Pro and $168 for Max on monthly billing, according to z.ai's subscribe page on 28 October 2026. Paying quarterly takes 20% off and paying yearly takes 30% off, which brings Lite to $12.60, Pro to $56 and Max to $117.60 a month. Prices are in US dollars.
| Plan | Monthly billing | Quarterly (20% off) | Yearly (30% off) | Usage vs Lite |
|---|---|---|---|---|
| Lite | $18/month | $14.40/month | $12.60/month | 1x (10,000 credits a week) |
| Pro | $80/month | $64/month | $56/month | 6x |
| Max | $168/month | $134.40/month | $117.60/month | 14x |
Earlier figures of $72 for Pro and $160 for Max, a 10% monthly discount, and Pro at 5x and Max at 20x Lite no longer match z.ai's page. Today Pro is $80 and 6x Lite, Max is $168 and 14x Lite, and monthly billing has no discount.
What Is the Difference Between Lite, Pro and Max?
The tiers differ in usage allowance and priority, not in which models you can use: all three support GLM-5.3 and GLM-5.3-Flash. Lite is aimed at lightweight work on small repositories, Pro at day-to-day development on mid-sized repositories with 6 times Lite's usage, and Max at advanced users on larger repositories with 14 times Lite's usage.
| Plan | 5-hour credits | Weekly credits | Extras |
|---|---|---|---|
| Lite | 2,000 | 10,000 | Latest flagship models, 20+ agent tools, default data privacy |
| Pro | 12,000 | 60,000 | Priority access to new models, curated MCP tools, faster generation |
| Max | 28,000 | 140,000 | First access to new models, dedicated resources during peak times |
How credits are counted
z.ai converts usage into credits with token multipliers: for GLM-5.3, input tokens count at 6.9, cached input at 1.7 and output at 24, divided by 10,000; GLM-5.3-Flash uses lower multipliers. Outside peak hours, Monday to Friday 14:00 to 18:00 Singapore time, model usage is billed at half the usual credit rate, so scheduling heavy jobs off-peak stretches your allowance.
Is the Coding Plan Cheaper Than the API?
For steady, daily coding, the plan is usually cheaper than paying per token in 2026. The GLM-5.3 API is priced at $1.40 per million input tokens and $4.40 per million output tokens, while the plan bundles a large allowance into a flat fee. z.ai estimates fully off-peak use comes out as much as 92% cheaper than paying per call for GLM-5.3.
| Option | Price | Best for |
|---|---|---|
| Coding Plan Lite | From $12.60 to $18 a month | Steady individual coding on smaller projects |
| Coding Plan Pro or Max | From $56 to $168 a month | Full-time AI-assisted development and agent sessions |
| GLM-5.3 API | $1.40 input, $4.40 output per 1M tokens | Occasional use, or GLM inside your own product |
| GLM-5.3-Flash API | $0.15 input, $0.50 output per 1M tokens | High-volume, lighter tasks |
From 16 October 2026, z.ai manages standard API keys and Coding Plan keys separately. A standard key used in a coding tool bills pay-as-you-go and does not use your plan quota, so create a Coding Plan key for your tools.
If you are comparing open-weight options beyond GLM, see our guides to DeepSeek V4.1 Flash and Qwen3.8.
Which Tools Work With the GLM Coding Plan?
z.ai lists support for 20+ agent tools in 2026, including ZCode, its own agent-powered IDE, plus Claude Code, Codex, OpenCode, Cline, Kilo Code, Cursor, AutoClaw and OpenClaw. Every plan includes vision understanding and the Web Search, Web Reader and Zread MCP tools, so agents can read screenshots, search the web and read repositories.
- ZCode: z.ai's own IDE, optimised for GLM models.
- Claude Code, Codex, OpenCode: terminal and agent tools configured with a Coding Plan key.
- Cline, Kilo Code, Cursor: editor-based coding assistants.
- AutoClaw and OpenClaw: agent tools for broader automation; see our OpenClaw guide.
For a small software studio in India or elsewhere, the maths is simple: if two developers use AI for most of the working day, per-token API costs can quickly exceed a Pro or Max subscription, so a plan with off-peak scheduling usually wins. Track a week of usage on Lite first to see where you land.
How Do You Choose the Right Tier in 2026?
Pick the tier by your real weekly usage. Start on monthly Lite or Pro, watch how quickly you hit the 5-hour and weekly limits for a couple of weeks, then move up or down. Switch to quarterly or yearly billing only once your volume is stable, since the 20% and 30% discounts reward commitment, not experimentation.
- Occasional or side-project coding: Lite.
- Full-time developer using AI most of the day: Pro.
- Constant agent sessions or large repositories: Max.
- Teams: the Team plan adds seat management, usage dashboards, optional overage billing and centralised invoicing.
For background on the models, see our guides to GLM-5.2 and GLM-5.3, and our GLM API integration guide.