Amazon SageMaker AI Instances Comparison
Discover and compare Amazon SageMaker machine learning instance specifications and pricing on a single page, covering various workload types and regions. Filter by GPU, inference accelerators, vCPUs, and memory. Compare prices across training, inference, and studio workloads.
The data is frequently updated from the AWS API, but may not be accurate. This site is not affiliated with Amazon or AWS. The latest update occurred on September 4, 2026 at 12:31 AM UTC
| Select | Instance Type | Instance Family | vCPUs | Memory (GiB) | GPUs | Instance Price | Best Alternative Instances | Savings Options | Best Price Region / Diff |
|---|---|---|---|---|---|---|---|---|---|
| ml.inf1.24xlarge | Accelerated Computing Instances | 48 | 192 | 0 | 6.9430 | -18% | |||
| ml.inf1.2xlarge | Accelerated Computing Instances | 4 | 16 | 0 | 0.5320 | -18% | |||
| ml.inf1.6xlarge | Accelerated Computing Instances | 12 | 48 | 0 | 1.7360 | -18% | |||
| ml.inf1.xlarge | Accelerated Computing Instances | 2 | 8 | 0 | 0.3640 | -18% | |||
| ml.g4dn.12xlarge | Accelerated Computing Instances | 24 | 192 | 4 | 5.9980 | -18% | |||
| ml.g4dn.16xlarge | Accelerated Computing Instances | 32 | 256 | 1 | 6.6730 | -18% | |||
| ml.g4dn.2xlarge | Accelerated Computing Instances | 4 | 32 | 1 | 1.1530 | -18% | |||
| ml.g4dn.4xlarge | Accelerated Computing Instances | 8 | 64 | 1 | 1.8460 | -18% | |||
| ml.g4dn.8xlarge | Accelerated Computing Instances | 16 | 128 | 1 | 3.3360 | -18% | |||
| ml.g4dn.xlarge | Accelerated Computing Instances | 2 | 16 | 1 | 0.9030 | -18% | |||
| ml.c5.18xlarge | Standard Instances | 36 | 144 | 0 | 4.5620 | -19% | |||
| ml.c5.2xlarge | Standard Instances | 4 | 16 | 0 | 0.5060 | -19% | |||
| ml.c5.4xlarge | Standard Instances | 8 | 32 | 0 | 1.0140 | -19% | |||
| ml.c5.9xlarge | Standard Instances | 18 | 72 | 0 | 2.2810 | -19% | |||
| ml.c5.large | Standard Instances | 1 | 4 | 0 | 0.1270 | -19% | |||
| ml.c5.xlarge | Standard Instances | 2 | 8 | 0 | 0.2530 | -19% | |||
| ml.c5d.18xlarge | Compute Optimized Instances | 36 | 144 | 0 | 5.1790 | -19% | |||
| ml.c5d.2xlarge | Compute Optimized Instances | 4 | 16 | 0 | 0.5760 | -19% | |||
| ml.c5d.4xlarge | Compute Optimized Instances | 8 | 32 | 0 | 1.1510 | -19% | |||
| ml.c5d.9xlarge | Compute Optimized Instances | 18 | 72 | 0 | 2.5900 | -19% | |||
| ml.c5d.large | Compute Optimized Instances | 1 | 4 | 0 | 0.1440 | -20% | |||
| ml.c5d.xlarge | Compute Optimized Instances | 2 | 8 | 0 | 0.2880 | -20% | |||
| ml.m5.12xlarge | Standard Instances | 24 | 192 | 0 | 3.3900 | -18% | |||
| ml.m5.24xlarge | Standard Instances | 48 | 384 | 0 | 6.7800 | -18% | |||
| ml.m5.2xlarge | Standard Instances | 4 | 32 | 0 | 0.5650 | -18% | |||
| ml.m5.4xlarge | Standard Instances | 8 | 64 | 0 | 1.1304 | -18% | |||
| ml.m5.large | Standard Instances | 1 | 8 | 0 | 0.1420 | -19% | |||
| ml.m5.xlarge | Standard Instances | 2 | 16 | 0 | 0.2820 | -18% | |||
| ml.m5d.12xlarge | Standard Instances | 24 | 192 | 0 | 3.9910 | -18% | |||
| ml.m5d.24xlarge | Standard Instances | 48 | 384 | 0 | 7.9840 | -18% | |||
| ml.m5d.2xlarge | Standard Instances | 4 | 32 | 0 | 0.6650 | -18% | |||
| ml.m5d.large | Standard Instances | 1 | 8 | 0 | 0.1670 | -18% | |||
| ml.m5d.xlarge | Standard Instances | 2 | 16 | 0 | 0.3320 | -18% | |||
| ml.r5.12xlarge | Standard Instances | 24 | 384 | 0 | 4.4664 | -18% | |||
| ml.r5.24xlarge | Standard Instances | 48 | 768 | 0 | 8.9340 | -18% | |||
| ml.r5.2xlarge | Standard Instances | 4 | 64 | 0 | 0.7440 | -18% | |||
| ml.r5.4xlarge | Standard Instances | 8 | 128 | 0 | 1.4892 | -18% | |||
| ml.r5.large | Standard Instances | 1 | 16 | 0 | 0.1860 | -20% | |||
| ml.r5.xlarge | Standard Instances | 2 | 32 | 0 | 0.3720 | -18% | |||
| ml.r5d.12xlarge | Memory Optimized Instances | 24 | 384 | 0 | 5.0690 | -18% | |||
| ml.r5d.24xlarge | Memory Optimized Instances | 48 | 768 | 0 | 10.1380 | -18% | |||
| ml.r5d.2xlarge | Memory Optimized Instances | 4 | 64 | 0 | 0.8450 | -18% | |||
| ml.r5d.4xlarge | Memory Optimized Instances | 8 | 128 | 0 | 1.6900 | -18% | |||
| ml.r5d.large | Memory Optimized Instances | 1 | 16 | 0 | 0.2110 | -18% | |||
| ml.r5d.xlarge | Memory Optimized Instances | 2 | 32 | 0 | 0.4220 | -18% | |||
| ml.c6g.12xlarge | Compute Optimized Instances | 48 | 96 | 0 | 2.4480 | -49% | |||
| ml.c6g.16xlarge | Compute Optimized Instances | 64 | 128 | 0 | 3.2640 | -49% | |||
| ml.c6g.2xlarge | Compute Optimized Instances | 8 | 16 | 0 | 0.4080 | -49% | |||
| ml.c6g.4xlarge | Compute Optimized Instances | 16 | 32 | 0 | 0.8160 | -49% | |||
| ml.c6g.8xlarge | Compute Optimized Instances | 32 | 64 | 0 | 1.6320 | -49% | |||
| ml.c6g.large | Compute Optimized Instances | 2 | 4 | 0 | 0.1020 | -49% | |||
| ml.c6g.xlarge | Compute Optimized Instances | 4 | 8 | 0 | 0.2040 | -49% | |||
| ml.m6g.12xlarge | Standard Instances | 48 | 192 | 0 | 2.7072 | -46% | |||
| ml.m6g.16xlarge | Standard Instances | 64 | 256 | 0 | 3.6096 | -46% | |||
| ml.m6g.2xlarge | Standard Instances | 8 | 32 | 0 | 0.4512 | -46% | |||
| ml.m6g.4xlarge | Standard Instances | 16 | 64 | 0 | 0.9024 | -46% | |||
| ml.m6g.8xlarge | Standard Instances | 32 | 128 | 0 | 1.8048 | -46% | |||
| ml.m6g.large | Standard Instances | 2 | 8 | 0 | 0.1128 | -46% | |||
| ml.m6g.xlarge | Standard Instances | 4 | 16 | 0 | 0.2256 | -46% |