Robbed by Z.AI? My usage numbers make absolutely no sense
I subscribed to Z.AI’s smallest Coding Plan on August 16th, two days ago, for $18.
My first experience was already pretty rough: I managed to hit the 2,000-credit 5-hour limit in about 20 minutes of coding. GLM stopped right in the middle of making changes and left the project in a non-compiling state.
Fair enough, I can accept that part. I’m responsible for planning around the limits of the plan I bought.
Today I let GLM continue the changes. The project compiles again, although I still need to go through everything and check whether some of the changes were destructive.
But then I looked at my usage dashboard, and this is where things stop making sense.
According to the usage graph:
- August 16: roughly 2,000 credits used
- Today: roughly 760 credits used
- Expected total: ~2,760 credits
So out of a weekly allowance of 10,000, I should have roughly 7,240 credits remaining.
Instead, the usage report says I only have:
3,250 / 10,000 remaining.
That means their system has apparently counted about 6,750 credits of usage.
So there are roughly 3,990 credits missing/unaccounted for more than my entire visible usage today and August 16 combined.
For context, I have used GLM 5.3 exclusively through OpenCode. I have not used the web interface for GLM.
Am I misunderstanding how Z.AI calculates Coding Plan credits? Are there hidden multipliers, retries, tool calls, cached/context tokens, or anything else that gets charged but isn't shown in the usage graph?
Because if the dashboard is supposed to represent my billed usage, the numbers simply do not add up.
Has anyone else experienced this with Z.AI’s Coding Plan?