atlas
g4dn…
Try

7 of 1173 instances · cheapest is g4dn.xlarge at $0.5870/hr

Instance Workload vCPUs Memory Local disk Network On-demand/hr ↑ Per vCPU
g4dn.xlargeGPU & accelerators 4 16 GiB 125 GB NVMe Up to 25 Gbps $0.5870 $0.1467
g4dn.2xlargeGPU & accelerators 8 32 GiB 225 GB NVMe Up to 25 Gbps $0.8380 $0.1047
g4dn.4xlargeGPU & accelerators 16 64 GiB 225 GB NVMe Up to 25 Gbps $1.342 $0.0839
g4dn.8xlargeGPU & accelerators 32 128 GiB 900 GB NVMe 50 Gbps $2.426 $0.0758
g4dn.12xlargeGPU & accelerators 48 192 GiB 900 GB NVMe 50 Gbps $4.362 $0.0909
g4dn.16xlargeGPU & accelerators 64 256 GiB 900 GB NVMe 50 Gbps $4.853 $0.0758
g4dn.metalGPU & accelerators 96 384 GiB 1.8 TB NVMe 100 Gbps $8.724 $0.0909