Amazon Neptune Pricing Calculator
Compare provisioned instances vs Serverless NCUs with Standard vs I/O-Optimized storage — and see when Neptune Analytics is cheaper than keeping graph compute running 24/7.
- No signup
- Instant results
- ~3 min
Rates reviewed
Cluster mode
us-east-1list rates, pricing as of 2026-07-05. Source: AWS public pricing.
Who This Tool Is For
Cloud architects and ML engineers designing knowledge graphs, fraud detection, or Bedrock RAG with graph retrieval.
Why We Built This Tool
Neptune bills compute, storage, and I/O separately — I/O-Optimized flips the math when I/O exceeds ~25% of total database spend.
What Problem It Solves
- I/O surprise. Standard storage charges $0.20/million I/O requests on top of instance cost.
- Serverless floor. 0.5 NCU minimum runs continuously unless you delete the cluster.
- Analytics vs Database. Batch graph analytics on m-NCU can beat 24/7 provisioned compute for periodic jobs.
See Generative AI on AWS or compare Bedrock pricing.
Frequently Asked Questions
When should I pick I/O-Optimized over Standard storage?
AWS recommends I/O-Optimized when I/O costs exceed 25% of total Neptune spend. At $0.225/GB-month vs $0.10/GB plus $0.20/M I/O, break-even is roughly 125M I/O requests per 100 GB stored per month.
What is the minimum Neptune Serverless NCU charge?
Neptune Serverless bills per NCU-second with a minimum capacity of 0.5 NCU. At $0.1098/NCU-hour in us-east-1, 0.5 NCU running 24/7 is about $40/month before storage.
How does Neptune Analytics m-NCU pricing differ from the database?
Analytics charges per m-NCU-hour while the graph job runs. Paused graphs bill at 10% of active compute. Database Serverless uses NCUs that scale with query load on the primary graph store.
Does Multi-AZ double my Neptune compute bill?
Multi-AZ doubles compute for provisioned instances and the primary serverless writer. Read replicas add their own instance or NCU charges on top.
How do Neptune costs compare to OpenSearch for RAG?
OpenSearch Serverless OCUs suit vector search; Neptune suits relationship-heavy graph retrieval. Model both in their respective calculators before committing architecture.
Where do these prices come from?
Rates are AWS published us-east-1 list prices (Amazon Neptune pricing), pricing as of 2026-07-05. Other regions vary ~5–15%.