Amazon Athena Pricing Calculator
Model on-demand scan volume and provisioned DPUs — see monthly cost before an unpartitioned table runs in production.
- No signup
- Instant results
- ~2 min
Rates reviewed
Read the full breakdown — Companion pricing guide covers billing dimensions, traps, and a 30-day cleanup plan.
Which billing mode applies?
us-east-1list rates, pricing as of 2026-06-26. DDL and failed queries are not charged. Result caching (7-day) not modeled. Source: AWS public pricing.
Who This Tool Is For
Data engineers and FinOps leads running ad-hoc SQL on S3, CUR 2.0 exports, or CloudTrail archives.
Why We Built This Tool
Athena has no idle cluster cost — but unpartitioned CSV tables turn every analyst query into a full-bucket scan at $5/TB.
What Problem It Solves
- Scan volume. Every dollar is proportional to bytes read — partition pruning is the primary lever.
- Provisioned vs on-demand. Steady high-volume workloads may need reserved DPUs; sporadic queries should stay on-demand.
- Workgroup guardrails. Per-query scan caps prevent runaway costs from unoptimized SQL.
Frequently Asked Questions
How does Amazon Athena pricing work?
On-demand queries bill $5 per TB of data scanned (rounded up to 10 MB minimum per query). DDL and failed queries are not charged.
When should I use provisioned capacity?
Provisioned capacity at ~$0.30/DPU-hour suits predictable, high-concurrency workloads. Sporadic analytics should stay on-demand to avoid idle DPU cost.
Does Athena result caching reduce cost?
Yes — identical queries against unchanged data within 7 days are served from cache at no charge. Caching is automatic.
How much can partitioning save?
Filtering to one month of a year partitioned by month reduces scan cost ~92%. Columnar Parquet adds another 70–95% vs CSV.
Athena vs CloudTrail Lake queries?
Lake queries bill $0.005/GB scanned; Athena on S3 archives bills $5/TB ($0.005/GB — same per-GB rate, different minimums). Compare with the CloudTrail calculator for ingestion cost.
Should I use workgroups for cost control?
Yes. Workgroups enforce per-query scan limits, separate result locations, and cost attribution per team via tags.