Free Tier

To get you started with AWS DeepComposer, we provide a Free Tier, so you can use Generative AI to compose new music without incurring any costs, for 12 months. You can choose a sample model for a musical genre (say, rock or pop), input a melody using the keyboard, and AWS DeepComposer performs machine learning inference in the cloud to generate an entirely new music composition, with 4-part accompaniment. Your free tier of up to 500 inference jobs lasts for 12 months and starts when you first use the AWS DeepComposer service.

Learn more about AWS Free Tier »

Free Trial

In addition to the Free Tier, we provide a 30-day AWS DeepComposer Free Trial to all users, so you can train your first Generative AI models to compose music at no cost. The Free Trial covers training 4 models and generating new musical compositions using those models up to 40 times, which should help you get started creating compositions.

Usage Based Pricing

AWS DeepComposer usage is billed based on an hourly rate as follows:



AWS DeepComposer - Training

$1.26 per hour

AWS DeepComposer - Inference

$2.14 per hour

Please see pricing example #1 below for a breakdown of costs you can expect for typical AWS DeepComposer training and music generation (also known as inference), after you have used the Free Tier and Free Trial.

Pricing example 1 (when outside the free trial)

This example assumes you have used all of your Free Trial, to provide a clear view of the costs of using the service. When using AWS DeepComposer, developers learn about how Generative AI works through training generative models, and then creating music through machine learning inference, in the cloud. The AWS DeepComposer console will guide you to create and train your first generative model; a typical training should take on average 8 hours, but could take more or less time depending on the parameters and data set selected. After training completes, you can execute inference against your new model to create music based on an input musical phrase. This music generation (or inference) typically takes about 1 minute, or .0167 hours.

In this example, you train a new model (1 training job of 8 hours) and then try 5 different input melodies to create 5 musical compositions with the newly trained model (5 inference requests). The AWS DeepComposer service bills customers directly; you will see the price for these activities as follows:

Service Consumed Hours Cost per hour Total ($)
AWS DeepComposer - Training 8 $1.26 $10.08
AWS DeepComposer - Inference 0.0083 $2.14 $0.18
Total     $10.26