Tiers & limits
What the project tier chip means, and the limits Tattva enforces on uploads, queries and sessions.
The tier chip
Settings › Project shows a tier chip next to the project's creation date. New projects are created as free. No feature or quota depends on the tier: every project can use everything described in these docs. There are no plans to choose and no in-app upgrade.
Limits Tattva enforces
| What | Limit |
|---|---|
| Uploaded file size | 200 MB per file. On deployments without cloud storage, 50 MB. |
Skill archive (.skill) | 5 MB |
| Rows the agent reads from an uploaded file per query | 10,000 by default |
| Rows the agent reads from BigQuery per query | 50,000 at most |
| Data one BigQuery query may scan | 20 GiB by default. BigQuery rejects a larger query before it runs, so it costs nothing. |
| Time one BigQuery query may run | 2 minutes by default |
| Data a stored metric definition may scan per run | 5 GiB (the same cap applies to Try it) |
| Runs kept per tracker question | The last 30 |
| Audit log on screen | The 50 most recent events |
| API key expiry | 1 to 3,650 days, or none |
| Sign-in session | 30 days, unless you sign out |
If a question needs more data than an uploaded file allows, load the data into BigQuery and connect it there. Check the Trail's Compute card when a figure looks too small: it shows the SQL that ran.
Limits set by your operator
Your platform operator controls the infrastructure underneath: BigQuery quotas and slots, database connections, storage. If you hit one of those, you'll see an error; pass it to your operator.
What isn't limited
Projects per workspace, sources per project, trackers, questions per day and API calls aren't capped by Tattva. Your operator's cloud quotas still apply.