AI budgets that do not surprise you
How REM AI keeps AI spend predictable - plan budgets, per-key limits, alerts before the money runs out, and usage you can explain line by line.
The worst AI bill is the one you find out about at the end of the month: a retry loop in a background job, a forgotten test key, a prompt that grew to fifty pages. REM AI is built so that this does not happen quietly. Here is how the pieces fit together.
1. A budget per organisation
Every plan comes with a monthly allowance of REM Credits and a budget for the period. Calls are charged at the provider's cost of each call, so a short question costs a fraction of a credit and a long document costs more. When the budget is reached, calls stop with a clear error until the period resets or the plan changes. Nothing keeps spending in the background.
2. Alerts before the limit
When the organisation's spend passes 80% of the budget, its owners get one e-mail from support@remplatforms.com, and the alert shows under Alerts. That is the moment to look at Usage grouped by key and see where it went.
3. Limits per key
On Pro and above, owners can give each AI key its own budget, budget period, requests and tokens per minute, parallel requests, and even per-model limits. A good pattern:
Key | Budget | Limits |
|---|---|---|
Production app | most of the budget | the plan's limits |
Background jobs | a small budget | low requests per minute |
Tests and contractors | a small budget and an expiry | low limits |
4. Rate limits that protect everyone
Requests and tokens per minute and parallel requests come from your plan. A runaway loop hits 429 long before it can empty the budget, and well-behaved clients retry with backoff.
5. Usage you can explain
Every call is in Usage and, for admins, in the Request Log: who or which key, which model, how many tokens, how many credits, whether it came from the cache. Tag keys by project and group by tag to see spend per project; set user on calls (Max) to see spend per end customer of your own app.
Start with the defaults
You do not need to set anything to be safe: the plan's budget, alerts and rate limits apply from the first call. Add per-key limits when you have more than one app, and look at Usage once a week.
Keep reading
Build an app by describing it
What happens between the sentence you type into REM Vibe Coder and the app that comes back - and how to write the sentence so you get the app you meant.