Amazon EC2 Per Second Billing
Amazon EC2 Elastic GPUs and Amazon EBS volumes will also move from per hour billing to per second billing with a minimum of 60 seconds. As with EC2 instances, the list prices will continue to be displayed as it is today in GB-month for capacity, IOPS-month provisioned for io1 IOPS and per Million I/O requests for Magnetic IOPS.
Today, customers may have use cases that require large number of instances running for irregular periods of time, such as dev/test, data processing, analytics, batch processing, image rendering and gaming applications. To improve flexibility and allow customers to focus on their application instead of maximizing usage to the hour, with EC2 per second billing, customers can start, stop, and terminate EC2 instances and take comfort in the fact that their EC2 resources will not be charged for the minutes and seconds remaining in the hours their instances were running.
This change is effective in all AWS Regions and is effective immediately, for Linux instances that are newly launched or already running. Per-second billing is not currently applicable to instances running Microsoft Windows or Linux distributions that have a separate hourly charge.
Read the entire article here, Announcing Amazon EC2 per second billing
Via the fine folks at Amazon Web Services.