Usage-based billing
The dimensions Silos meters — compute time, storage, data transfer, and branch/replica hours — and how scale-to-zero changes the math.
Silos meters usage rather than charging for provisioned capacity. You pay for the resources your databases actually consume, on top of your plan.
Pricing is evolving and subject to change. This page describes what is metered; for current rates and details, contact us.
What's metered
Compute time
Time a database's runtime is actually serving queries. Idle databases scale to zero and incur no compute charge.
Storage
The durable state your databases hold — snapshots and WAL — in object storage.
Data transfer
Data moved in and out, including reads served to clients and replicas.
Branch & replica hours
Time that branches and edge replicas are materialized and consuming resources.
Why scale-to-zero matters for cost
The defining property of the model is that compute is only billed while a database is serving traffic. An idle database has no runtime to pay for — you carry only its storage cost. This is what makes patterns like a database-per-tenant or ephemeral databases for agents economical: dozens or hundreds of mostly-idle databases cost far less than the same number of always-on servers.
The trade-off is the cold start: the first query after a period of inactivity pays to materialize the database. For latency-critical paths, keep a database warm.
Seeing your usage
Your usage is visible in the Console's billing and usage views, and through the REST API:
curl "https://api.silos.sh/v1/projects/{project_id}/usage" \
-H "Authorization: Bearer $SILOS_API_KEY"Use these to understand which dimension drives your bill and to set spending controls accordingly.