Is AWS GPU Instances free? What it actually costs in 2026
No free tier. It is a paid product.
On-demand H100 capacity on AWS runs around $12.29 per GPU-hour, against roughly $1.99 at RunPod and $1.87 on the Vast.ai marketplace — a five-to-six-fold difference for the same silicon. Savings Plans and Reserved Instances cut 30–50% in exchange for commitment, and Spot capacity cuts more in exchange for interruption. Data egress is billed separately and is the line most people forget when they model a move. Prices checked 2026-07-30; GPU pricing moves faster than almost anything else in cloud, so confirm current rates before you plan around them.
When paying is still the right call
If your data already lives in S3, moving it out costs real money in egress and may cost more than the GPU savings — run that arithmetic before anything else. Stay too if you need enterprise compliance paperwork, a VPC that never touches the public internet, guaranteed capacity under contract, or tight integration with the AWS services around the model. For regulated workloads the specialist clouds often simply cannot supply the attestations, and that ends the discussion regardless of price.
What you are locked into
Lower than most AWS services, which is the good news. A container running CUDA is portable, so the model code usually moves unchanged. What holds you are the things around it: training data sitting in S3 with egress charges to move it, IAM and VPC wiring, and any dependency on SageMaker, Bedrock or EKS orchestration built around the instances rather than inside them.
If you would rather not pay: RunPod
RunPod
H100s from about $1.99/hr, running in under a minute.
Per-second billing, no commitment. H100 around $1.99/hr, with cheaper Community Cloud capacity and a pricier Secure Cloud tier. Serverless scales to zero, so idle inference endpoints cost nothing. Verified 2026-07-30.
See all 4 AWS GPU Instances alternatives comparedCompare the free options head-to-head
Pricing is verified against the vendor's published figures and dated above. Vendors change prices — confirm current numbers with Amazon Web Services before you commit. More cloud gpu & ai compute decisions on Macrostack.