I know they are stating publicly no refunds for token plan, but for first time users, do they make exception? On a day 1 without coding reaching weekly %28 usage is insane. Do they ever reply to their emails?
I think I found the rate limit issue. GLM5.1 is seriously drifts to rabbit hole digging like menace which burns the limit like maniac. This is purely GLM5.1 model issue not user issue.
it's an autonomous model, designed to execute non stop but taxing for users who is paying for usage. One would say this is equivalent to directly burning money on the porch.
If you have project with long dependencies and deep level architecture it will dig the shit out of it and snitch back to usage tracking and hit you back with rate limit. So even buying 20x plan won't help you.
Literally confused right now, first time tried GLM5.1 with $72 pro plan today, I have codex 20x so all coding done with GPT5.5 High. Used GLM5.1 on hermes with project overview and research, planning. One day Got rate limited. Weekly 27% use already. Finally was happy found something reliable over lazy opus4.6/4.7 but rate limit is insane. Another turn off was no direct image comprehension like MCP had to be used. Other than that GLM5.1 is impressive so far. I will pick GLM5.1 any day over Claude like fr. Seems this scam rate limit is widespread issue.