Pay for queries, not for idle.
Pay for the queries you run — usage-based on SaaS, or a platform fee for BYOC and on-prem.
SaaS
Free tier: first 2 query-hours/month. No minimums.
- Pay for query-seconds actually consumed
- No idle compute, no reserved warehouses
- Free tier for evaluation — 2 query-hours/month
- Full monitoring & cost estimates included
Your cloud (BYOC)
Data stays in your account.
- Runs in your AWS/GCP account
- Data never leaves your VPC or bucket
- Everything in SaaS, plus dedicated support
- Migration support included
On-prem
Contact us.
- Full platform on your hardware
- S3-compatible MinIO storage
- Built for air-gapped & regulated environments
- Annual license, custom terms
You keep the $6,500.
Questions, answered
How are agent queries billed?
Same as human queries: per query-second consumed. Budgets cap exposure; blocked queries cost nothing.
Does the MCP server cost extra?
No. MCP access is included in every tier.
What's the billing granularity?
Query-seconds. Each query runs as its own right-sized job, and you're billed for the CPU and memory it actually consumed — not for a warehouse size class or an always-on cluster.
What counts as usage?
Compute time for queries you run, whether against your lake (Iceberg/Parquet) or federated against Postgres and MySQL. Storage stays in your own bucket or databases, so it's never part of our bill.
How is Turbo mode priced?
Turbo allocates more compute for a faster result, so it costs more per query than Auto mode. You choose the trade-off per query, per user or per workspace — nothing switches to Turbo without you asking for it.
How do estimates relate to the final bill?
The pre-execution estimate is computed from the same signals — schema, index usage and logical data size — that determine the actual allocation, so it tracks closely with what you're billed. Estimates are a guide, not a cap; guardrails exist if you want a hard stop instead.
Do you help with migration?
Yes. Design partners and BYOC/on-prem customers get hands-on migration support — running your workload in parallel with your existing warehouse while you validate results and cut over on your own timeline.
Not sure which plan fits?
Tell us about your workloads and we'll recommend a setup — cloud, BYOC or on-prem.