How credits work
What costs credits, what is free, and why two drawing checks can cost very different amounts.
Credits are the fair-use unit. Most of what you do in Bimiq is free; only the actions that run an AI model over your documents are metered.
What is free
Everything that is not an AI run. In particular:
- Uploading documents, and importing them from a connected source
- Splitting a drawing set into sheets, and viewing the sheets
- Indexing documents so they can be searched
- Viewing reports, findings, issues and results
- Logging a decision by hand
- Raising an RFI by hand
- Grouping raw clashes into issues
That last one matters more than it looks. You can import a large clash report and see how many real issues it collapses into without spending anything.
What is metered
| Action | Cost |
|---|---|
| Project Intelligence message | 1 credit |
| Drawing check | 6 credits per sheet |
| Submittal review | 1 credit per page, minimum 10 |
| Clash triage | 1 credit per 8 issues, capped at 100 |
| Model vs Spec Check | 60 credits per run, per document |
| Document coordination | 100 credits per run |
| RFI assistance | 15 credits per RFI |
| Decision extraction | 15 credits per run |
The three that scale with the work
Most actions cost a flat amount. Three do not, and these are the ones that surprise people.
Drawing checks are per sheet, and uncapped
A drawing check runs a vision model over every page. A 40-sheet set costs 240 credits. A 400-sheet set costs 2,400.
There is no cap, deliberately. Every extra sheet is real work, so a cap would mean charging the same for a set ten times the size.
Splitting the set is free, so you always know the sheet count before you commit to the analysis.
Submittal reviews are per page, with a floor
One credit per page, but never less than 10. The floor exists so that a two-page cut sheet still covers the fixed cost of setting up the review.
Clash triage is per issue, and capped
Charged on grouped issues, not on raw clashes, at one credit per 8 issues. It stops at 100 credits.
The cap is there because only the top 800 issues get AI enrichment, so beyond that point the real cost plateaus. A run with 5,000 issues costs the same as a run with 800.
Why the same action cost different amounts
Almost always because the input was a different size. Check the sheet count, the page count, or the issue count.
The other reason is retrying. Each run is charged, so analysing the same set twice costs twice.
Before a run starts
Bimiq checks you can afford a run before starting it. Where the size is not yet known, it assumes a size rather than optimistically letting a run start that cannot finish. This is why a pre-flight estimate can be higher than the final charge.
When you run out
Metered actions are blocked when the balance is exhausted. Everything free stays available: your documents, findings, reports and history are not locked behind the balance. Nothing already produced is taken away.
Keeping usage down
- Check the sheet count before analysing. Splitting is free, so the number is there before you spend anything.
- Do not analyse a set twice. If you re-upload a revised set, analyse the revision, not the original again.
- Ask Project Intelligence freely. At 1 credit a message it is the cheapest thing here, and it often answers a question that would otherwise mean a full run.
- Import BCF and look at the issue count first. Grouping is free; only triage is charged.