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 | 2 credits |
| Drawing check | 6 credits per sheet, 8 when checked against a specification |
| Submittal review | 1 credit per page + 1 per 25 spec requirements checked, minimum 10 |
| Deep cross-discipline check | 2 credits per comparison |
| 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 four that scale with the work
Most actions cost a flat amount. Four 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.
Checking against a specification costs 8 a sheet instead of 6. Before each review you choose what the set is checked against: your codes and firm standards, a project specification, both, or neither. Standards are the base rate, because they are short documents you reuse across every project. A specification is hundreds of pages of requirements written for one job, and it is consulted on every sheet, so it carries 2 extra credits per sheet. The estimate shown before you start reflects whatever you have selected.
Unchanged sheets in a reissued set are not charged twice. Bimiq fingerprints every page when it splits a set. When you upload a later issue of the same drawings, any page that is byte-for-byte identical to one already reviewed in that project reuses the earlier reading instead of paying to read it again.
This matters because a revision usually changes a handful of pages out of hundreds, and reading the sheets is most of what a drawing check costs. A 250-sheet set is 1,500 credits the first time. If the next issue changes 30 sheets, it is about 180, not 1,500.
Two things to know:
- It matches on the page image, not the sheet number. A sheet that was redrawn, re-plotted at a different size, or re-exported by different software counts as changed even if it looks the same, and will be read again.
- It works across sets within a project, so it does not matter which upload came first or how the files were named.
The completed check tells you how many sheets were reused.
Submittal reviews are per page and per requirement
A review does two things, and each is charged for what it costs.
Reading the submittal is one credit per page.
Checking it against the spec is one credit per 25 requirements. This is the part people do not expect, because it has nothing to do with how long the submittal is. A three-page product datasheet checked against a section with 400 requirements is 400 requirements of work — the same as if the submittal had been two hundred pages.
A review is never less than 10 credits.
The completed review shows how many requirements were checked. That number is the one to look at when a review costs more than you expected — not the page count.
BIMIQ works out which spec section a submittal belongs to by reading it. When it can identify the section, it checks against that section's requirements only. When it cannot — an unusual submittal, or a spec whose sections are not clearly labelled — it deliberately checks against everything rather than risk scoping to the wrong section and passing a submittal that should have failed. That is the safe choice, and it is the expensive one, which is why the requirement count is worth a glance.
Deep cross-discipline checks are per comparison
The standard drawing check compares grid labels across disciplines. The optional deep check sends the actual drawings to the vision model in pairs — the architectural and structural plan of Level 2, side by side — and asks whether they are coordinated.
That is two full drawings per comparison, so it is charged at 2 credits each. The number of comparisons is shown before you start.
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.
For submittal reviews there is a second size that is easy to miss: how many spec requirements were checked. Two reviews of the same three-page datasheet can differ several-fold if one matched a spec section and the other had to be checked against the whole spec. The requirement count is shown on the completed review.
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. To carry on, buy a credit top-up from your billing settings, which is available immediately, or wait for your plan's credits to renew at the start of the next cycle.
Everything free stays available: your documents, findings, reports and history are not locked behind the balance. Nothing already produced is taken away.
Top-up credits are separate from your plan's included credits. They are not tied to the billing cycle, and they stay available for up to a year.
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 first. At 2 credits a message it is still 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.