AWS prices its services on a pay-as-you-go basis. You pay for what you use, and then layer in other pricing models, like On-Demand, Spot Instances, Reserved Instances, and Savings Plans, when you want lower rates for more predictable usage. Each service also has its own pricing dimensions, EC2 by compute time, S3 by storage, retrieval, and transfer, and Lambda by request count and execution duration.
If you want a quick estimate before you deploy, the AWS Pricing Calculator helps you model expected costs. For ongoing visibility and control after deployment, AWS cost optimization tools help you monitor, analyze, and manage spend over time. Below is a service-by-service breakdown of what AWS charges for and how the main pricing models apply.
| Service | Primary Pricing Basis |
|---|---|
| EC2 | Compute capacity (On-Demand, Reserved, Spot) |
| S3 | Storage usage, data retrieval, and transfer |
| EBS | Volume type and storage size |
| RDS | Instance type, storage, and data transfer |
| Bedrock | Input and output tokens processed |
| Lambda | Number of requests and execution duration |
| EFS | Storage consumption and data transfer |
This is part of an extensive series about cloud storage.
Editor’s note: Updated the article to cover recent market trends, updated product information to reflect features and capabilities in 2026.
Related Content:
Before we dive into pricing for specific Amazon services, here are the pricing models AWS offers. Understanding each one is the first step toward choosing the right fit for your workloads and not paying more than you need to.
|
Model |
Best For |
Discount Range |
Commitment Required |
Key Trade-off |
|---|---|---|---|---|
|
Free Tier |
New users exploring AWS |
Up to $200 in credits |
None |
6 months or until credits run out; limited service access |
|
On-Demand |
Unpredictable workloads |
None (baseline pricing) |
None |
Highest cost per hour |
|
Spot Instances |
Interruptible batch jobs |
Up to 90% |
None |
Can be terminated with 2-min notice |
|
Reserved Instances |
Steady-state workloads |
Up to 75% |
1-3 years |
Less flexibility to change |
|
Savings Plans |
Consistent usage patterns |
Up to 72% |
1-3 years |
Committed spend regardless of usage |
|
Dedicated Hosts |
Compliance/licensing needs |
Varies |
On-demand or reserved |
Higher cost, single-tenant only |
The AWS Free Tier now includes a Free plan with up to $200 in credits over 6 months, $100 available when you create an account and up to $100 more as you explore AWS. Your account closes automatically after 6 months or when the credits run out, and you're never charged unless you choose to convert to a Paid plan. AWS also offers short-term trials and always-free services with usage caps that do not expire.
On-demand pricing allows users to purchase compute and database capacity by the hour or second, depending on the service. There are no long-term commitments or upfront payments required, making it a highly flexible option. This model is suitable for applications with unpredictable workloads or for those that cannot be interrupted.
For example, if an application experiences variable traffic, users can quickly scale up or down the resources in response to demand. The pay-as-you-go approach ensures that users only pay for what they use, providing a cost-effective solution for short-term, spiky, or temporary workloads.
Spot instances offer the ability to bid for unused AWS capacity at significantly lower prices than on-demand instances. This pricing model is suitable for applications that have flexible start and end times, can tolerate interruptions, or are stateless and can be easily restarted. Spot instances can provide savings of up to 90% compared to on-demand prices.
They are suitable for batch processing, big data analytics, containerized workloads, CI/CD, and high-performance computing. However, it’s important to design applications to handle interruptions, as AWS can terminate spot instances with a two-minute warning when the capacity is needed elsewhere. This requires building fault-tolerant systems that can gracefully handle the loss of instances.
Reserved instances (RIs) provide a discount (up to 75%) compared to on-demand pricing in exchange for committing to use AWS resources for a one- or three-year term. RIs are suitable for applications with steady-state usage or predictable workloads. They come in three payment options: All Upfront (payment for the full commitment period made in advance), Partial Upfront, and No Upfront.
Additionally, AWS offers flexibility with convertible RIs, which can be exchanged for another RI with different attributes if business needs change. This model helps in achieving cost predictability and optimization, making it a strategic choice for long-term projects.
Savings plans are a flexible pricing model that provides significant savings on AWS usage, similar to Reserved Instances but with more flexibility. By committing to a consistent amount of usage (measured in USD per hour) for a one- or three-year term, users can save up to 72% on their AWS bills.
Savings Plans automatically apply to any usage that matches the plan, regardless of instance family, region, operating system, or tenancy. There are two types of Savings Plans: Compute Savings Plans, which apply to any EC2 instance usage, and EC2 Instance Savings Plans, which provide the most savings but apply only to specific instance families in a chosen region.
Dedicated hosts provide physical servers that are dedicated to a single customer, allowing them to use their existing server-bound software licenses and address the relevant compliance requirements. This model is suitable for users needing a single-tenant infrastructure for security, regulatory, or performance reasons.
Dedicated hosts give customers more control over their server placement and allow them to use their own licensing, which can result in cost savings. For example, organizations running legacy software that is licensed per physical server or core can take advantage of dedicated hosts to optimize their costs and meet licensing requirements. These hosts also support bringing your own license (BYOL) for software from vendors such as Microsoft and Oracle.
Flat-rate plans bundle multiple AWS services into one price with no overage charges, designed for simple, predictable monthly billing.
For services like S3 and data transfer out from EC2, the more you use, the less you pay per GB. Data transfer in is always free.
AWS pricing can vary significantly based on the region where services are deployed. Each AWS region operates independently, and prices can differ due to various factors, including the cost of electricity, real estate, and local taxes in each location.
US East (N. Virginia) tends to offer the lowest prices, but choosing a region is not only about cost. You also need to weigh latency, data residency and compliance requirements, and which AWS services are available in the region you want to use.
If you run workloads across multiple regions, those price differences add up faster, especially for compute-heavy services. Tracking spend by region is one of the first steps in AWS cost management because it shows you where architecture decisions are quietly increasing your bill.
Source: CloudPrice.net
The pricing details here are for the US East (N. Virginia) Region, and correct as of the time of this writing. For up-to-date pricing and additional details refer to the official AWS pricing page.
EC2 pricing depends on the instance family, size, region, operating system, and pricing model you choose.
On-Demand pricing examples in US East (N. Virginia) for Linux instances:
For EC2, Compute Savings Plans apply across instance families, sizes, operating systems, and regions, and can reduce costs by up to 66%. EC2 Instance Savings Plans are more restrictive because they are tied to a specific instance family in a specific region, but they can offer up to 72% off. Both options require a 1- or 3-year hourly commitment.
Spot Instances can reduce EC2 costs by up to 90%, but AWS can reclaim that capacity with two minutes' notice. They are best suited for interruptible workloads like batch processing, CI/CD jobs, data analysis, and other work that can tolerate interruption.
Dedicated hosts provide physical servers for exclusive use. On-demand pricing examples include:
Dedicated hosts are suitable for workloads requiring dedicated hardware, compliance, and licensing requirements.
Learn more in our detailed guide to AWS EC2 costs.
The key components of S3 pricing include storage costs, request and retrieval charges, data transfer fees, and additional features like data management and replication.
Source: AWS S3 Pricing
Amazon S3 storage costs vary based on the storage class and the amount of data stored. The primary storage classes include:
Users are charged for the requests made to S3 and the retrieval of data from specific storage classes:
In addition, there are special retrieval costs for data in infrequent access or archival storage classes:
AWS RDS pricing consists of several key components:
Source: AWS RDS Pricing
Instance Hourly Price
The hourly cost of an AWS RDS instance varies based on the instance type and the region. There are various instance types, each designed for different performance needs. For example:
You can choose between two pricing models:
AWS RDS supports three types of storage:
Data transfer costs apply to data moving in and out of AWS data centers. These costs vary based on the amount of data and the region. The main categories include:
Amazon Bedrock pricing varies by model, region, service tier, and access method, including on-demand, provisioned throughput, and batch. Effective per-token cost can also shift based on the model you choose, the size of the context window, and whether repeated prompts benefit from cache hits.
Tokens are only part of the total Bedrock bill. Agent runtime, Guardrails, knowledge base retrieval, and model evaluation can all add charges outside per-token pricing, which is why you need visibility into those supporting costs alongside model invocation if you want an accurate view of what AI features actually cost. For more on that, see FinOps for AI.
Finout gives you a unified AWS cost view with MegaBill, so you can see EC2, S3, RDS, Lambda, EKS, EBS, and Bedrock spend in one place instead of stitching reports together across accounts and services.
With Virtual Tags, you can allocate 100% of AWS spend to the right teams, environments, or business units without changing native AWS tags. CostGuard then surfaces idle resources, rightsizing opportunities, and commitment recommendations from day one, so optimization work starts with clear ownership.
Anomaly Detection helps you catch cost spikes before they turn into budget overruns, with alerts delivered through Slack or email. If you want faster answers, Billy lets you ask natural-language questions about spend and returns chart-backed answers using your live Finout data.
Finout's MCP server and Cost/Usage API let you bring governed cost data into developer tools, AI agents, and your existing data workflows. That gives teams one reliable cost context instead of separate exports and one-off scripts.
FinOps Agents help investigate anomalies, map ownership, and route optimization work through Jira, Slack, or ServiceNow. If your team is trying to stay lean while covering more spend, that is the difference between seeing waste and actually getting it fixed.
Learn more about Finout for AWS cost optimization
Together with our content partners, we have authored in-depth guides on several other topics that can also be useful as you explore the world of cloud storage.
Authored by Cloudian
Authored by Cloudian
Authored by Cloudinary