Understand usage and spend in Nomic
Nomic's pricing has two recurring components plus a small one-time cost per file:
| What | When you pay | Driver |
|---|---|---|
| Workflow spend | Every time someone runs a workflow or asks Assistant | Input size, output size, model used |
| Indexing spend | Once per file version, when the file is indexed | Pages parsed |
| Model usage | Rolled up into workflow spend | Choice of model per run |
Admins see all of this on Admin → Usage.
What's in "workflow spend"?
A workflow run pays for:
- Reading the inputs — the drawing set, specs, context files, and instructions.
- Thinking — the model's reasoning steps.
- Writing the output — the issue table, summary, and markups.
Bigger inputs → bigger spend. A 200-page drawing set costs more than a 5-page sample, roughly proportional to the page count.
What's in "indexing spend"?
Indexing is billed per page parsed, at $0.012 per page — a 100-sheet drawing set is about $1.20. It happens once per file version, and after that the file is searchable at no extra cost. Re-uploading a new version costs another indexing pass for that version.
Uploading or syncing a file does not necessarily index it. PDFs and legacy Office files index on upload; everything else, including files brought in by a sync, is indexed when someone runs Index Document or Index Folder. See Syncing vs indexing.
Why large jobs cost more than they "look like" they should
Two things commonly surprise teams:
- Fanout on multi-page documents. A 100-page drawing review isn't one model call — it's many. Most of the cost lives in the per-page work, not the summary.
- Context files multiply with the input. A workflow with 200 pages of standards still loads those standards on every run. If a standard isn't actually needed for the task, removing it from context saves spend.
- Dense files can behave like many pages. Large spreadsheets, scanned drawings, and table-heavy PDFs may take more work to parse than their file count suggests.
- Repeated uploads create repeated indexing. If the same large document is uploaded as a new file or new version, Nomic may need to index it again.
Read the dashboards
- Admin → Usage shows organization included usage for the period plus every AI transaction, with per-day and per-user totals.
- Admin → Analytics → Projects breaks spend and activity down by project, including last used time, members, attached files, workflows, and Assistant sessions. Admins can open a project detail page or export the filtered project table to CSV.
- Admin → Analytics → Workflows breaks spend down by workflow — useful for finding the expensive ones. Admins can export the filtered workflow table to CSV for spreadsheet analysis.
- Admin → Analytics → Parsing shows indexing spend by user and individual parse events.
- Admin → Members shows per-user spend against each member's limit, which helps identify heavy users who may need coaching, more headroom, or workflow support.
- Use available CSV exports when you need to analyze spend outside Nomic.
Estimate before a large run
If you need a budget estimate before running a large review, start with a representative sample instead of the full package:
- Pick a few pages or one small folder that reflects the real job.
- Run the workflow with the same standards, model, and output format you plan to use.
- Check the spend for that sample in Admin → Usage.
- Scale the estimate by the size of the full package, then add a buffer for dense sheets, spreadsheets, scanned pages, or extra verification steps.
This is especially useful during pilots, when admins are still learning which workflows and document types drive the most spend.
Reduce spend without losing quality
- Test workflows on a small sample first.
- Trim context files to what the task actually needs.
- Use the smaller model option for early drafting; switch to the bigger model for the final pass.
- Schedule large reviews after-hours so they don't compete with interactive use.
- Reuse indexed project files instead of uploading the same file into each workflow run.
- Use Assistant to find the relevant pages or sections before running a full workflow.