The platform runs on your accounts, so the main costs are the ones you’d expect:
| Cost |
Where it shows up |
| BigQuery queries (data processed) |
Your Google Cloud bill. See Activity for an estimate by person and app, and Query size and limits to see a query’s size before you run it. |
| BigQuery storage |
Your Google Cloud bill. Pipelines add raw tables; dbt adds models. |
| AI usage (Agent) |
Your own AI provider account. The monthly cap pauses the agent when reached. |
| Limit |
Default |
Notes |
| Largest BigQuery query in Explore and the Agent |
10 GB |
Admins change it in Project settings → Agent. |
| Agent monthly spend |
$20 |
Estimated from tokens; admins change it. |
| Largest BigQuery query in a Notebook |
100 GB billed |
Per query; can be overridden in code. |
| Notebook idle time before it stops |
60 minutes |
Files in notebooks/ are kept. |
| A Knowledge document |
200,000 characters |
Up to 500 documents per project. |
| Rows shown in an Agent step |
200 |
The agent sees the first 50 when reasoning, and it can aggregate for you. |
| Agent steps per question |
8 |
Ask something narrower if it runs out. |
| Activity page |
100 rows per page |
|
- Filter on date columns; select only what you need.
- Use aggregated models for dashboards many people open.
- Set the query limit to what your largest legitimate query needs, not higher.
- Check Activity for the heaviest queries and who runs them.