Gemini CLI vs Z.ai: tracking usage on Mac and Windows
This is not an argument about which tool writes better code - that depends on your work, and anyone claiming otherwise is guessing. It is about something narrower and more answerable: how each one is metered, and what you can actually see while you use them.
How each is tracked
Gemini CLI uses an existing local login, and usage is read locally, so it keeps working when the network does not.
Z.ai uses a Z.ai API key, and readings come from the provider, so a network problem shows as a stale number rather than a wrong one.
How Gemini CLI and Z.ai meter your usage
Gemini CLI is no longer something you get with a consumer Google plan. Google ended Gemini CLI service for Google AI Pro, Google AI Ultra and the free Gemini Code Assist for individuals on June 18, 2026, while Gemini Code Assist Standard and Enterprise customers and paid API keys keep access (Google Developers Blog). The Gemini CLI quota page now opens with a notice that, for unpaid tier and Google One users, Gemini CLI was replaced by Antigravity CLI.
Licensed seats are counted in requests, not tokens: up to 1,500 model requests per user per day on Standard and 2,000 on Enterprise, pooled with Code Assist agent mode, and a single prompt can use several of them (Google Cloud quota reference). A paid API key is billed per token; Gemini 3.8 Flash costs $0.75 per million input tokens and $3.75 per million output tokens through December 31, 2026, rising to $1.50 and $7.50 from January 1, 2027 (Gemini API pricing). On the API, requests-per-day quotas reset at midnight Pacific time, and depending on your billing history a spend-based limit is also checked over a rolling 10-minute window (rate limit docs). In a session, /stats model shows token usage alongside your quota limits (Gemini CLI docs).
Z.ai sells the GLM Coding Plan, a coding subscription starting at 18 USD a month in Lite, Pro and Max tiers, for tools such as Claude Code, Cline and OpenCode (Z.ai plan overview). Every tier has two credit windows: Lite gets 2,000 credits per 5 hours and 10,000 per week, Pro 12,000 and 60,000, and Max 28,000 and 140,000. The 5-hour credits come back 5 hours after they are spent, and the weekly credits reset every 7 days counted from when you subscribed. Credits are worked out from tokens with per-model multipliers, so GLM-5.3 output costs three times as many credits as GLM-5.3-Flash output, and model usage outside peak hours (Monday to Friday, 14:00 to 18:00 Singapore time) is charged at half the credit rate.
Quota progress is on the Z.ai subscription page, and a Charge Type page lists tokens by pricing type and tool calls (Z.ai plan overview). When the plan quota is used up, you wait for the next 5-hour cycle: Z.ai says it will not deduct from your account balance, and plan subscribers cannot make API calls outside the plan (Z.ai FAQ).
The difference shows up in pacing. A Gemini license hands you one daily request budget, while Z.ai refills in rolling 5-hour slices under a weekly ceiling and charges less outside its peak hours, so when you work changes how far a Z.ai plan goes.
Plan details checked against each vendor's own pages on October 3, 2026.
Why this matters if you run both
Running two tools means two places to check, two sets of limits, and two chances to be surprised mid-task. The cost is rarely the money - it is losing your thread because something stopped answering and you did not see it coming.
Watching both in one place
SessionWatcher shows Gemini CLI and Z.ai together in the Mac menu bar or the Windows taskbar. It reads and displays usage on your Mac or PC; it does not read your code or your conversations. That boundary matters more when you are pointing one tool at two providers.
Per-tool detail is in the Gemini CLI tracker guide and the Z.ai tracker guide.
Frequently asked questions
Can I track both at once?
Yes - both appear in the same reading, in the Mac menu bar or the Windows taskbar.
What does each need?
Gemini CLI uses an existing local login. Z.ai uses a Z.ai API key.
Does it read my code?
No. Usage only, never code or conversation content.
Can I use the GLM Coding Plan inside Gemini CLI?
Z.ai limits the plan to the tools on its supported list, which includes Claude Code, Codex, OpenCode, Cursor and Cline, and says use in unsupported tools may result in restricted benefits (Z.ai usage policy). Gemini CLI was not on that list when we checked on October 3, 2026.
Does Z.ai bill my balance when the plan runs out, the way a Gemini API key keeps billing?
No. Z.ai says GLM calls in supported tools only use the Coding Plan quota, so when it runs out you wait for the next 5-hour refresh (Z.ai FAQ). A paid Gemini API key bills every token at the published rates, so what stops you there is a rate limit such as requests per day, which resets at midnight Pacific time (Gemini API rate limits).
SessionWatcherRunning both?
Watch them in one menu bar.
Native app for the Mac menu bar and the Windows taskbar. Track Claude, Codex and 20 more tools' usage, costs, and rate limits in real time.


“Fast, simple, and does exactly what it should. Definitely worth it.”
@nicojerome on GitHub
macOS 14+. Free for 7 days, no credit card.





