.png)
Key takeaways
1
Neptune has two products with distinct billing models: Neptune Database, which runs as Provisioned or Serverless, and Neptune Analytics.
2
Database Savings Plans can cut eligible Neptune costs by up to 35% for a one-year hourly commitment with no upfront payment. Coverage includes eligible provisioned instance usage, Neptune Serverless, and Neptune Analytics.
3
Under Standard storage, I/O charges can exceed instance cost on heavy-traversal workloads. Consider I/O-Optimized when I/O exceeds roughly 25% of total Neptune database spend and a configuration-specific comparison supports it.
AWS Neptune pricing starts at $0.348/hr for a db.r5.large on-demand in US East, but the instance rate is rarely the whole story. Neptune spans two products and three pricing options, each billed differently, and depending on your workload, storage or I/O can end up costing more than compute. This guide breaks down every cost component and where it actually lands in your bill.
The short answer
Neptune’s total cost depends less on the instance you pick than on whether you run Neptune Database in Provisioned or Serverless mode, or Neptune Analytics, because each has a different billing model. I/O-heavy workloads can shift the largest cost from compute to storage and I/O.
The discount path follows the same logic: AWS Database Savings Plans apply to eligible Neptune provisioned instance usage, Serverless, and Analytics. Right-size or modernize a cluster when the performance or price-performance case supports it, then size the commitment to stable eligible hourly spend.
The discount path follows the same logic: AWS Database Savings Plans apply to eligible Neptune provisioned instance usage, Serverless, and Analytics. Right-size or modernize a cluster when the performance or price-performance case supports it, then size the commitment to stable eligible hourly spend.
AWS Neptune: Two Products, Three Pricing Options
Neptune Database is the core managed graph database for OLTP workloads: social graphs, fraud detection, recommendation engines, knowledge graphs, and network topology. It supports Gremlin and openCypher (property graph) and SPARQL (RDF).
It runs as Provisioned instances or as Neptune Serverless, on shared cluster storage that grows automatically in 10 GB increments.
Neptune Analytics is a separate in-memory engine for graph algorithms PageRank, community detection, shortest path, vector similarity search covered in its own section below.
AWS prices Neptune three ways: on-demand, Serverless, and Database Savings Plans.
Discount coverage has expanded quickly:
It runs as Provisioned instances or as Neptune Serverless, on shared cluster storage that grows automatically in 10 GB increments.
Neptune Analytics is a separate in-memory engine for graph algorithms PageRank, community detection, shortest path, vector similarity search covered in its own section below.
AWS prices Neptune three ways: on-demand, Serverless, and Database Savings Plans.
Discount coverage has expanded quickly:
- December 2025: Database Savings Plans launch; Neptune Database covered from day one
- March 5, 2026: coverage extends to Neptune Analytics
Neptune Database Instance Pricing: Current Generation Rates
| Instance | Generation / RAM | On-Demand/hr | Performance & Notes |
|---|---|---|---|
| db.r5.large | Intel x86 · 16 GB | $0.348 | Baseline. Previous generation, widely referenced. |
| db.r5.xlarge | Intel x86 · 32 GB | $0.696 | Baseline. Double the r5.large rate. |
| db.r6g.large | Graviton2 · 16 GB | ~$0.313 | ~10% lower than r5. ARM-based. |
| db.r7i.large | Intel Xeon Gen4 · 16 GB | Verify at Amazon Neptune pricing | Up to +15% price-performance vs R6i. Launched March 2025. Requires Neptune engine v1.4.3+. |
| db.r7g.large | Graviton3 · 16 GB | ~$0.263 | ~24% lower than r5. Launched May 2025. Requires Neptune engine v1.4.5+. |
| db.r8g.large | Graviton4 · 16 GB | ~$0.263 | ~24% lower than r5; up to 30% better performance than R7g. Launched May 2025. Requires Neptune engine v1.4.5+. |
All rates in this table are for US East (N. Virginia). The R5 and I/O-Optimized example rates are published by AWS; the R7g and R8g estimates apply AWS’s stated 16% discount versus comparable R6g rates. Check the Amazon Neptune pricing page for current Regional rates.
Graviton4 R8g: What the 4.7x Means
AWS launched Graviton3-based R7g and Graviton4-based R8g instances for Neptune on May 1, 2025. Both are priced 16% lower than equivalent R6g instances and require Neptune engine version 1.4.5 or higher; AWS documents the upgrade as an instance-type modification.
The performance gain is significant. AWS Database Blog benchmarks show up to 4.7x better write and 3.7x better read query price-performance for v1.4.5 on R8g versus previous versions, a combined effect of the engine upgrade and the Graviton4 hardware.
For write-heavy graph workloads, that’s a measurable improvement at a lower per-instance rate.
The performance gain is significant. AWS Database Blog benchmarks show up to 4.7x better write and 3.7x better read query price-performance for v1.4.5 on R8g versus previous versions, a combined effect of the engine upgrade and the Graviton4 hardware.
For write-heavy graph workloads, that’s a measurable improvement at a lower per-instance rate.
For the full migration path, region availability, and sequencing, see our AWS Neptune guide.
Neptune Storage Configuration: Standard vs I/O-Optimized
In US East (N. Virginia), Neptune Standard storage is $0.10/GB-month, with I/O billed separately at $0.20 per million requests. Rates vary by Region in Frankfurt, for example, AWS prices the same components at $0.119/GB-month and $0.22 per million I/O.
Example: a 500 GB database with 100 million I/O requests per month costs $50 storage + $20 I/O = $70/month in storage and I/O charges.
Neptune I/O-Optimized storage at $0.225/GB-month with no I/O charges, but instance-hour rates are meaningfully higher: a db.r5.large runs $0.4698/hr under I/O-Optimized versus $0.348 Standard, roughly a 35% premium.
The break-even, per AWS: if your I/O spend exceeds 25% of your total Neptune database spend, I/O-Optimized can save up to 40% overall.
Example: a 500 GB database with 100 million I/O requests per month costs $50 storage + $20 I/O = $70/month in storage and I/O charges.
Neptune I/O-Optimized storage at $0.225/GB-month with no I/O charges, but instance-hour rates are meaningfully higher: a db.r5.large runs $0.4698/hr under I/O-Optimized versus $0.348 Standard, roughly a 35% premium.
The break-even, per AWS: if your I/O spend exceeds 25% of your total Neptune database spend, I/O-Optimized can save up to 40% overall.
Standard
=
$254 instance
+
$50 storage
+
$200 I/O
=
~$504/month
I/O-Optimized
=
$343 instance
+
$112.50 storage
=
~$455/month
I/O-Optimized saves ≈ $49/month here and the crossover shifts with cluster size, since the instance premium applies to every node.
The I/O trap for graph workloads: traversal-heavy queries that follow edges across many hops generate far more I/O per query than typical relational queries. A friends-of-friends query within 3 hops may traverse thousands of edges, thousands of storage I/O requests per execution.
Before choosing, total 30 days of the CloudWatch VolumeReadIOPs and VolumeWriteIOPs metrics, then price the billed I/O at the applicable Regional rate.
If I/O spending exceeds roughly 25% of total Neptune database spend, compare Standard and I/O-Optimized costs for the full cluster configuration.
The I/O trap for graph workloads: traversal-heavy queries that follow edges across many hops generate far more I/O per query than typical relational queries. A friends-of-friends query within 3 hops may traverse thousands of edges, thousands of storage I/O requests per execution.
Before choosing, total 30 days of the CloudWatch VolumeReadIOPs and VolumeWriteIOPs metrics, then price the billed I/O at the applicable Regional rate.
If I/O spending exceeds roughly 25% of total Neptune database spend, compare Standard and I/O-Optimized costs for the full cluster configuration.
Neptune Serverless: NCU Pricing and Use Cases
Neptune Serverless scales automatically between your configured minimum and maximum NCU bounds; each NCU provides approximately 2 GB of memory with associated CPU and networking.
Billing is per NCU-second at $0.1098/NCU-hour in US East, from 1 to 128 NCUs. At 1 NCU running continuously, that’s about $80/month minimum; storage and I/O are billed separately at Standard rates.
To size a baseline: minimum NCUs × active hours × the regional NCU rate, then add Standard storage and I/O. Two NCUs held continuously work out to 2 × 730 × $0.1098 ≈ $160/month before storage.
Serverless can suit variable workloads, development environments, and new applications because capacity scales within configured NCU bounds.
Compare observed NCU usage, minimum capacity, storage, I/O, and an equivalent provisioned configuration before selecting a long-running production model.
For the full NCU break-even methodology, see Amazon Neptune Serverless Pricing: NCU Costs & Break-Even.
Billing is per NCU-second at $0.1098/NCU-hour in US East, from 1 to 128 NCUs. At 1 NCU running continuously, that’s about $80/month minimum; storage and I/O are billed separately at Standard rates.
To size a baseline: minimum NCUs × active hours × the regional NCU rate, then add Standard storage and I/O. Two NCUs held continuously work out to 2 × 730 × $0.1098 ≈ $160/month before storage.
Serverless can suit variable workloads, development environments, and new applications because capacity scales within configured NCU bounds.
Compare observed NCU usage, minimum capacity, storage, I/O, and an equivalent provisioned configuration before selecting a long-running production model.
For the full NCU break-even methodology, see Amazon Neptune Serverless Pricing: NCU Costs & Break-Even.
Neptune Analytics: Separate Product, Different Cost Model
Neptune Analytics keeps graphs in memory for analytical algorithms across the full graph, and is billed in m-NCUs each bundling 1 GB of memory with compute and network capacity for one hour.
Rates vary by region and are published on the Amazon Neptune pricing page; use the AWS Pricing Calculator to model your graph size and job frequency.
The pause capability is the key economic lever: while paused, you pay approximately 10% of the normal compute price, with all data and settings retained. Resuming is a single API call.
Database Savings Plans have covered Neptune Analytics since March 5, 2026. For scheduled jobs, compare pause-resume operating time with expected eligible hourly usage before sizing a commitment to stable spend.
Rates vary by region and are published on the Amazon Neptune pricing page; use the AWS Pricing Calculator to model your graph size and job frequency.
The pause capability is the key economic lever: while paused, you pay approximately 10% of the normal compute price, with all data and settings retained. Resuming is a single API call.
Database Savings Plans have covered Neptune Analytics since March 5, 2026. For scheduled jobs, compare pause-resume operating time with expected eligible hourly usage before sizing a commitment to stable spend.
Neptune Backup and Additional Cost Components
Backup storage beyond the free 100% allowance
CPU credit overage on T4g/T3 instances above baseline
Each read replica billed at full instance rate
Data transfer between AZs and services
Backup storage is free up to 100% of your database storage size, then ~$0.021/GB-month. Long retention windows (30+ days) on write-heavy databases can accumulate 1–3 TB of backup data $21–63/month above the free tier.
T4g/T3 CPU credits: these instances run in Unlimited mode; sustained CPU above the instance baseline can incur charges at $0.15 per vCPU-hour. AWS positions these constrained instance types for getting started, development, and testing rather than production performance testing.
Read replicas: up to 15 per cluster, each billed at the full instance rate. Storage is shared, so only compute multiplies, a five-node db.r8g.large cluster runs ≈ $960/month on-demand (5 × $0.263 × 730).
Data transfer: transfer in is free, and the first 100 GB/month transferred out is free. Same-AZ EC2-to-Neptune traffic and Multi-AZ replication traffic are free; cross-AZ EC2 charges depend on whether the Neptune instance sits inside or outside a VPC, so confirm the applicable architecture in AWS pricing.
Global Database: replicated write I/Os between the primary Region and each secondary Region bill on top of standard instance, storage, cross-Region transfer, and backup rates.
Neptune workbench: notebooks are hosted and billed through Amazon SageMaker, priced per notebook-instance hour in the Ready state, independent of which Neptune engine you run.
Free trial: new Neptune customers get 30 days of a t3/t4g.medium instance (750 hours), 10M I/O, 1 GB storage, and 1 GB backup; there is no always-free tier.
T4g/T3 CPU credits: these instances run in Unlimited mode; sustained CPU above the instance baseline can incur charges at $0.15 per vCPU-hour. AWS positions these constrained instance types for getting started, development, and testing rather than production performance testing.
Read replicas: up to 15 per cluster, each billed at the full instance rate. Storage is shared, so only compute multiplies, a five-node db.r8g.large cluster runs ≈ $960/month on-demand (5 × $0.263 × 730).
Data transfer: transfer in is free, and the first 100 GB/month transferred out is free. Same-AZ EC2-to-Neptune traffic and Multi-AZ replication traffic are free; cross-AZ EC2 charges depend on whether the Neptune instance sits inside or outside a VPC, so confirm the applicable architecture in AWS pricing.
Global Database: replicated write I/Os between the primary Region and each secondary Region bill on top of standard instance, storage, cross-Region transfer, and backup rates.
Neptune workbench: notebooks are hosted and billed through Amazon SageMaker, priced per notebook-instance hour in the Ready state, independent of which Neptune engine you run.
Free trial: new Neptune customers get 30 days of a t3/t4g.medium instance (750 hours), 10M I/O, 1 GB storage, and 1 GB backup; there is no always-free tier.
On-Demand vs. Database Savings Plans: Which to Choose
Neptune has one commitment path, and what it covers determines the sequence you follow:
| Factor | Database Savings Plans |
|---|---|
| Max discount | Up to 35% |
| Term options | 1-year only |
| Upfront payment | None |
| Covered usage | Gen 7+ provisioned instances, Serverless, and Analytics (Analytics since March 5, 2026) |
| Flexibility | Applies regardless of supported engine, instance family, size, deployment option, or Region |
| Usage above the commitment | Billed at on-demand rates |
| Availability | All AWS Regions except China Regions |
Sources: AWS Database Savings Plans pricing page and Savings Plans FAQs, AWS announcements Dec 2, 2025 and Mar 5, 2026.
The short rule: evaluate whether a move from R5 or R6g to an eligible newer-generation instance improves price-performance for your workload before sizing a Database Savings Plans commitment.
If you migrate, complete the instance-type change and observe stable eligible hourly usage before committing to the sustained spend floor. Serverless can qualify for the program’s highest stated discount tier.
Anything above or outside the commitment keeps billing at on-demand rates, which is why the commitment is sized to the stable floor rather than to peak.
For program-wide mechanics, see the AWS Database Savings Plans complete guide.
The short rule: evaluate whether a move from R5 or R6g to an eligible newer-generation instance improves price-performance for your workload before sizing a Database Savings Plans commitment.
If you migrate, complete the instance-type change and observe stable eligible hourly usage before committing to the sustained spend floor. Serverless can qualify for the program’s highest stated discount tier.
Anything above or outside the commitment keeps billing at on-demand rates, which is why the commitment is sized to the stable floor rather than to peak.
For program-wide mechanics, see the AWS Database Savings Plans complete guide.
How Usage.ai Optimizes Neptune Costs
Manual commitment analysis works for a single cluster. It becomes much harder when Neptune spans Provisioned, Serverless, and Analytics deployments alongside RDS, EC2, and ElastiCache, and workloads continue to evolve.
We analyze your usage data and recommend qualifying commitments sized to stable eligible hourly spend.
With Flex Insured Commitments, teams can get up to 35% savings on a 1-year Neptune Database Savings Plan with none of the commitment risk.
If eligible usage drops and a qualifying Neptune Flex Commitment ends up costing more than the on-demand rate for that usage, we provide cashback protection on that difference, subject to program terms.
Our fee is a percentage of realized savings only. If we don’t save you money, you don’t pay anything.
We analyze your usage data and recommend qualifying commitments sized to stable eligible hourly spend.
With Flex Insured Commitments, teams can get up to 35% savings on a 1-year Neptune Database Savings Plan with none of the commitment risk.
If eligible usage drops and a qualifying Neptune Flex Commitment ends up costing more than the on-demand rate for that usage, we provide cashback protection on that difference, subject to program terms.
Our fee is a percentage of realized savings only. If we don’t save you money, you don’t pay anything.
Frequently asked questions
What does AWS Neptune cost?
Provisioned instances start at $0.348/hr (db.r5.large, US East); Serverless starts at $0.1098/NCU-hour (~$80/month minimum). Storage, I/O, backups, and replicas bill separately.
What are the discount options for AWS Neptune?
Database Savings Plans are the commitment discount available for Neptune: up to 35% in exchange for a consistent hourly commitment over a one-year term with no upfront payment, covering Gen 7+ provisioned instances, Neptune Serverless, and Neptune Analytics (since March 5, 2026). Usage above or outside the commitment bills at on-demand rates.
How is Neptune Analytics priced?
In m-NCUs 1 GB memory plus compute per hour, at region-specific published rates. Paused graphs cost ~10% of normal compute while retaining all data.
What is the difference between Neptune Standard and I/O-Optimized?
In US East: Standard is $0.10/GB-month plus $0.20 per million I/O; I/O-Optimized is $0.225/GB-month with I/O included but higher instance rates. Switch when I/O exceeds ~25% of total Neptune spend.
Does Neptune have a free tier?
A 30-day free trial only: 750 hours of t3/t4g.medium, 10M I/O, 1 GB storage, 1 GB backup. After that, standard on-demand rates apply.
Share
Facebook
X
LinkedIn
Reddit
