Important Update
Thank you for your interest in Amazon Elastic Inference. Amazon Elastic Inference is no longer available to new customers. You can get better performance at lower cost for your machine learning inference workloads by using other hardware acceleration options such as AWS Inferentia. If you are currently using Amazon Elastic Inference, please consider migrating your workload to these alternatives. To learn more, visit AWS Machine Learning Infrastructure page.
With Amazon Elastic Inference, you pay only for the accelerator hours you use. There are no upfront costs or minimum fees. There is no charge for the AWS-optimized versions of the TensorFlow and Apache MXNet deep learning frameworks. There are no additional charges for AWS PrivateLink VPC Endpoints to Amazon Elastic Inference, as long as you have at least one instance configured with an accelerator, that is running in an Availability Zone where a VPC endpoint is provisioned.
Amazon Elastic Inference accelerator pricing for Amazon EC2
Following is the Amazon Elastic Inference pricing with Amazon EC2 instances and Amazon ECS. For Amazon Elastic Inference pricing with Amazon SageMaker instances, please see the Model Deployment section on the Amazon SageMaker pricing page.
We have 2 families of Elastic Inference Accelerators with 3 different types in each. EIA2 is our latest accelerator type with twice the GPU memory of equivalent EIA1 accelerators. You can attach EIA2 to any EC2 instance.
Note: Please check Regional availability for the 2 accelerator families as these might differ.
-
US East
-
US West
-
EU
-
Asia Pacific
-
US East
-
US East (N. Virginia) Region
Accelerator type Throughput in trillion 32-bit floating point operations per second (FP-32 TFLOPS) Throughput in trillion 16-bit floating point operations per second (FP-16 TFLOPS) Memory Pricing eia2.medium 1 FP-32 TFLOPS 8 FP-16 TFLOPS 2 GB $0.120 per hour eia2.large 2 FP-32 TFLOPS 16 FP-16 TFLOPS 4 GB $0.240 per hour eia2.xlarge 4 FP-32 TFLOPS 32 FP-16 TFLOPS 8 GB $0.340 per hour eia1.medium 1 FP-32 TFLOPS 8 FP-16 TFLOPS 1 GB $0.130 per hour eia1.large 2 FP-32 TFLOPS 16 FP-16 TFLOPS 2 GB $0.260 per hour eia1.xlarge 4 FP-32 TFLOPS 32 FP-16 TFLOPS 4 GB $0.520 per hour US East (Ohio) Region
Accelerator type Throughput in trillion 32-bit floating point operations per second (FP-32 TFLOPS) Throughput in trillion 16-bit floating point operations per second (FP-16 TFLOPS) Memory Pricing eia2.medium 1 FP-32 TFLOPS 8 FP-16 TFLOPS 2 GB $0.120 per hour eia2.large 2 FP-32 TFLOPS 16 FP-16 TFLOPS 4 GB $0.240 per hour eia2.xlarge 4 FP-32 TFLOPS 32 FP-16 TFLOPS 8 GB $0.340 per hour eia1.medium 1 FP-32 TFLOPS 8 FP-16 TFLOPS 1 GB $0.130 per hour eia1.large 2 FP-32 TFLOPS 16 FP-16 TFLOPS 2 GB $0.260 per hour eia1.xlarge 4 FP-32 TFLOPS 32 FP-16 TFLOPS 4 GB $0.520 per hour -
US West
-
US West (Oregon) Region
Accelerator type Throughput in trillion 32-bit floating point operations per second (FP-32 TFLOPS) Throughput in trillion 16-bit floating point operations per second (FP-16 TFLOPS) Memory Pricing eia2.medium 1 FP-32 TFLOPS 8 FP-16 TFLOPS 2 GB $0.120 per hour eia2.large 2 FP-32 TFLOPS 16 FP-16 TFLOPS 4 GB $0.240 per hour eia2.xlarge 4 FP-32 TFLOPS 32 FP-16 TFLOPS 8 GB $0.340 per hour eia1.medium 1 FP-32 TFLOPS 8 FP-16 TFLOPS 1 GB $0.130 per hour eia1.large 2 FP-32 TFLOPS 16 FP-16 TFLOPS 2 GB $0.260 per hour eia1.xlarge 4 FP-32 TFLOPS 32 FP-16 TFLOPS 4 GB $0.520 per hour -
EU
-
EU (Ireland) Region
Accelerator type Throughput in trillion 32-bit floating point operations per second (FP-32 TFLOPS) Throughput in trillion 16-bit floating point operations per second (FP-16 TFLOPS) Memory Pricing eia2.medium 1 FP-32 TFLOPS 8 FP-16 TFLOPS 2 GB $0.129 per hour eia2.large 2 FP-32 TFLOPS 16 FP-16 TFLOPS 4 GB $0.258 per hour eia2.xlarge 4 FP-32 TFLOPS 32 FP-16 TFLOPS 8 GB $0.366 per hour eia1.medium 1 FP-32 TFLOPS 8 FP-16 TFLOPS 1 GB $0.140 per hour eia1.large 2 FP-32 TFLOPS 16 FP-16 TFLOPS 2 GB $0.280 per hour eia1.xlarge 4 FP-32 TFLOPS 32 FP-16 TFLOPS 4 GB $0.560 per hour -
Asia Pacific
-
Asia Pacific (Tokyo) Region
Accelerator type Throughput in trillion 32-bit floating point operations per second (FP-32 TFLOPS) Throughput in trillion 16-bit floating point operations per second (FP-16 TFLOPS) Memory Pricing eia2.medium 1 FP-32 TFLOPS 8 FP-16 TFLOPS 2 GB $0.205 per hour eia2.large 2 FP-32 TFLOPS 16 FP-16 TFLOPS 4 GB $0.411 per hour eia2.xlarge 4 FP-32 TFLOPS 32 FP-16 TFLOPS 8 GB $0.582 per hour eia1.medium 1 FP-32 TFLOPS 8 FP-16 TFLOPS 1 GB $0.220 per hour eia1.large 2 FP-32 TFLOPS 16 FP-16 TFLOPS 2 GB $0.450 per hour eia1.xlarge 4 FP-32 TFLOPS 32 FP-16 TFLOPS 4 GB $0.890 per hour Asia Pacific (Seoul) Region
Accelerator type Throughput in trillion 32-bit floating point operations per second (FP-32 TFLOPS) Throughput in trillion 16-bit floating point operations per second (FP-16 TFLOPS) Memory Pricing eia2.medium 1 FP-32 TFLOPS 8 FP-16 TFLOPS 2 GB $0.196 per hour eia2.large 2 FP-32 TFLOPS 16 FP-16 TFLOPS 4 GB $0.392 per hour eia2.xlarge 4 FP-32 TFLOPS 32 FP-16 TFLOPS 8 GB $0.556 per hour eia1.medium 1 FP-32 TFLOPS 8 FP-16 TFLOPS 1 GB $0.210 per hour eia1.large 2 FP-32 TFLOPS 16 FP-16 TFLOPS 2 GB $0.430 per hour eia1.xlarge 4 FP-32 TFLOPS 32 FP-16 TFLOPS 4 GB $0.850 per hour
Pricing example 1
Let’s say you are running a streaming video analytics application. To run deep learning inference to analyze a single video stream in this application, you can choose an Amazon EC2 c5.xlarge instance configured with an Amazon Elastic Inference eia2.medium accelerator. Your hourly cost to run this deep learning model in the US East (N.Virginia) region is:
Hourly price of c5.xlarge instance: $0.17
Hourly price of a eia2.medium accelerator: $0.12
Total hourly price of the instance configured with the accelerator: $0.29
Total monthly cost = $0.29 * 24 * 31 = $215.76
Pricing example 2
Let’s say you are running a web application that analyzes images uploaded by your end users in real time. To use deep learning inference for this application, you can choose an Amazon EC2 c5.large instance configured with an Amazon Elastic Inference eia2.medium accelerator and scale this instance capacity using Amazon EC2 Auto Scaling to meet the demands of your application. Your hourly cost for this combination in the US East (N.Virginia) region is:
Hourly price of c5.large instance: $0.085
Hourly price of a eia2.medium accelerator: $0.12
Total hourly price of the instance configured with the accelerator: $0.205
Total monthly cost = $0.205 * 24 * 31 = $152.52
Additional pricing resources
Easily calculate your monthly costs with AWS
Contact AWS specialists to get a personalized quote
Get started with Amazon Elastic Inference on Amazon SageMaker or Amazon EC2.