Z.ai Works from Russia Code and terminal
ZCode
The official GLM-5.3 harness, open source under Apache 2.0; the GLM Coding Plan meters token-based credits per 5 hours and per week.
Price and access
- Free?
- No, paid plans only
- Price
- Lite $18 / Pro $80 / Max $168 a month; 30% off yearly
- From Russia
- Works from Russia by our check; the pricing page lists no country limits.
- Models
- GLM-5.3 · GLM-5.3-Flash
- Licence
- Apache 2.0
- Our benchmark
- Models in our benchmark →
Official pricing page ↗ Checked 3 November 2026.
What changed in price and access
- Z.ai made its flagship GLM-5-Turbo model fully closed ↗
- Z.ai split its keys: a standard API key bills the balance, only a Coding Plan key draws on plan quota ↗
- ZCode's source code was opened on GitHub under the Apache 2.0 licence ↗
- Traffic analysis: while a user is logged in, ZCode silently uploads the workspace, including the full git history, to Alibaba Cloud storage ↗
- GLM 5.3 shipped: 1M-token context, API at $1.4 / $4.4 per million; ZCode runs on it by default ↗
What it gives a manager
- Team plan: an admin assigns and reassigns seats, sees usage per member and sets per-member caps on extra spend.
- Tasks can be sent from a phone via Telegram, WeChat or Feishu, and goal mode plans, executes and verifies the work on its own.
- Spend is predictable: 5-hour and weekly limits, and off-peak (outside 14:00–18:00 UTC+8 on weekdays) credits burn at half the rate.
- One subscription works not only in ZCode but also in Claude Code, Codex and OpenCode, so the team is not tied to one harness.
Weak points
- Per Tokenstead's analysis, a logged-in ZCode uploads the workspace with its git history to Alibaba Cloud; settings do not stop it.
- Subagents are weak: project memory skips them, 3 of Superbash's 16 projects would not run, and 40% of a weekly Lite limit went in a day.
- GLM-5.3 cannot see images itself: a separate Vision Understanding MCP tool reads screenshots, not the model in the chat.
- Past roughly 200k tokens in one thread the output visibly degrades; you have to open a new thread and re-summarise.
How to check this agent’s work
- Do not open folders with other people's data or secrets in ZCode: while you are logged in it uploads the workspace with its git history.
- Check the key: only a Coding Plan key draws on the plan quota, a standard key bills the balance.
- Check the result with a second model from another vendor.
Picking an agent takes an evening. After that the question is not the price but whether you can trust what it did.
You’re on the list
We will write when enrolment opens. Meanwhile, here is what the course covers.
The short course programmeWhat the course covers: programme and price
Strengths
- Credits are token-based and cached input is 4 times cheaper than fresh input: Lite gives 2,000 credits per 5 hours and 10,000 a week.
- Off-peak calls (anything outside 14:00–18:00 UTC+8 on weekdays) cost half the points, which covers almost the whole US and EU workday.
- The app keeps the GLM cache above 98%; AICodeKing puts that at roughly 30% more effective tokens than a third-party harness.
- Open source under Apache 2.0, with goal mode, skills, MCP, hooks, a built-in browser and launching via Telegram, WeChat or Feishu.
Alternatives
Kilo Code
Kimi Code CLI
DeepSeek HarnessWho it is for
- Team on Next.js that needs an inexpensive model with access from Russia
- Manager who needs an inexpensive team agent with seat assignment and spend limits
Example from a review: An uptime-monitoring MVP on the free tier
Shcoder used Plan mode with GLM-5.3 at high reasoning effort to design a service that pings sites every minute and reports uptime as percentages with charts, then handed the build to the cheaper GLM-5.3 Flash. The service worked on the first pass: sites could be added, checks ran, uptime was computed. The agent skipped the incidents section the first time and had to be asked again; the dark theme was fine, the light theme came out rough. He deployed it publicly; the video does not state how many free-tier tokens the project consumed.
Source: Shcoder, '100 МИЛЛИОНОВ токенов бесплатно?! Тестирую новый GLM 5.3 в ZCode' (in Russian)
Reviews (4)
The video authors’ opinions, not our conclusions.
- GLM-5.3 + Z-Code (New Free Tier & BEST WAYS): This is the CHEAPEST GOOD CODER RIGHT NOW! · AICodeKing, 18 August 2026In the reviewer's view, Lite at $12.60 with caching, off-peak hours and goal mode gives a Claude-class model, and he struggles to hit the Lite limits.
- GLM 5.3 Review (Don't Use Subagents on ZCode) · Superbash (BoxminingAI), 17 August 2026The reviewer rates the model highly but ZCode's subagents poorly: let GLM-5.3 write the code itself rather than delegate.
- 100 МИЛЛИОНОВ токенов бесплатно?! Тестирую новый GLM 5.3 в ZCode · Shcoder, 12 September 2026The reviewer recommends trying it: the free pool covers a small MVP, and he finds both the subscription and the API cheap.
- Best AI agent harnesses and how I use them (pi, omp, Zcode) · 0xSero, 2 September 2026The reviewer calls ZCode the desktop app closest to Codex, with loops, automations and computer use built in, and uses it daily.
Facts with sources
- The system prompt is almost entirely environment restrictions: don't start the dev server, don't run a build. Source (in Russian)
- Claude Code + GLM-5.3 on Terminal-Bench: 41.82% of tasks solved for $2.7K, versus 51.82% for $6.0K with Opus 5. Source (in Russian)
- Z.ai access isn't restricted from Russia. Source (in Russian)
- GLM Coding Plan monthly prices: Lite $18, Pro $80, Max $168; yearly billing takes 30% off, giving $12.60, $56 and $117.60. Source
- ZCode's source is on GitHub under Apache 2.0; the latest release, 3.17.1 of October 30, ships for macOS and Windows. Source
- Since October 16, API keys and Coding Plan keys are separate: a standard key bills your balance even when pointed at the Coding Plan endpoint. Source
