← All instances

g5.12xlarge

⚡ 4x GPU • 96.0 GB

Part of the G5 Accelerated Computing family — Hardware accelerators (GPUs, FPGAs) for specialized compute.

Name:gGPU Graphics & ML Inferenceg = GPU Graphics & ML InferencePowered by GPUs (NVIDIA/AMD) for graphics-intensive applications, video encoding, remote workstations, and machine learning inference.55th Generation5 = 5th GenerationAWS 5th generation hardware platform with updated CPU/system architecture.12xlarge12 Extra Large12xlarge = 12 Extra Large48 vCPUs • 192.0 GiB RAM • 3800 GB NVMe SSD • 40 Gigabit Network • 4x NVIDIA A10G (96.0 GB VRAM)
Price (Linux)
$5.6720/ hr
~$4,140.56/mo
Compute
48vCPUs
24 cores • x86_64
Memory
192.0GiB
4.0 GiB / vCPU
Value
$0.1182/ vCPU-hr
$0.0295/GiB-hr
Network
40Gbps
Baseline: 40 Gbps
Storage & EBS
2,000MB/s
Max: 65,000 IOPS

Technical Specifications & Limits

Hardware specs, container thresholds, and system limits

Compute & Processor

ProcessorAMD EPYC 7R32
Architecturex86_64 (x86_64)
vCPUs / Threads48 vCPUs
Physical Cores24
Clock Speed2.8 GHz
Intel AVX / AVX2No
Intel Turbo BoostNo
Bare MetalNo (Virtualized)

Performance & Benchmarks

CoreMark Score895,147
CoreMark / vCPU18,648.9 / core
FFmpeg Transcoding109 FPS
CoreMark per Dollar157,819 score / $
Cost per vCPU-hr$0.1182
Cost per GiB-hr$0.0295

Containers & Kubernetes

EKS Max Pods737 pods
Max ECS Tasks120 tasks
Max ENIs15
IPv4/v6 per ENI50
Total Addressable IPs750 total IPs
CNI Prefix DelegationSupported (High Density)

Networking & Bandwidth

Network Performance40 Gigabit
Baseline Bandwidth40 Gbps
Burst / Peak Bandwidth40 Gbps
Enhanced NetworkingYes
EFA SupportNo
IPv6 SupportNative Dual-Stack Supported

Storage & EBS Optimization

Storage ConfigurationNVMe SSD (3800 GB)
Local NVMe Disk3800 GB
EBS Baseline Throughput2000 MB/s
EBS Max Throughput2000 MB/s
EBS Baseline IOPS65,000
EBS Max / Burst IOPS65,000
Dedicated EBS Throughput16000 Mbps

GPU & Hardware Acceleration

GPU ModelNVIDIA A10G
Accelerator Count4x GPU(s)
Total GPU Memory96.0 GB
Memory ArchitectureGDDR / HBM
CUDA Compute Capabilityv8.6
Acceleration RuntimesCUDA, TensorRT, cuDNN, PyTorch, PyJAX

Platform & Architecture Details

Hypervisor / SystemAWS Nitro System (PCIe Offload)
Nitro EnclavesSupported (Isolated security enclaves)
Normalization Factor5.638170974
Generation StatusCurrent Generation (Recommended)
Date Introduced2021-11-12

Regional Pricing Matrix

Live on-demand rates across all global AWS cloud regions
Lowest Price Region
$5.6720/ hr
US East (N. Virginia) (us-east-1)
Highest Price Region
$9.6413/ hr
South America (Sao Paulo) (sa-east-1)
Global Fleet Average
$6.9702/ hr
Across all active regions
Regional Coverage
19regions
Available worldwide
Linux — Available in all 19 regions
Region NameRegion CodeOn-Demand Price
Asia Pacific (Hong Kong)
ap-east-1$8.7324
Asia Pacific (Jakarta)
ap-southeast-3$7.9385
Asia Pacific (Mumbai)
ap-south-1$6.8109
Asia Pacific (Seoul)
ap-northeast-2$6.9744
Asia Pacific (Sydney)
ap-southeast-2$7.3747
Asia Pacific (Tokyo)
ap-northeast-1$8.2261
Canada (Central)
ca-central-1$6.2978
EU (Frankfurt)
eu-central-1$7.0928
EU (Ireland)
eu-west-1$6.3317
EU (London)
eu-west-2$7.1999
EU (Paris)
eu-west-3$7.1999
EU (Stockholm)
eu-north-1$6.0159
Europe (Spain)
eu-south-2$5.9765
Israel (Tel Aviv)
il-central-1$6.6482
Middle East (UAE)
me-central-1$6.9575
South America (Sao Paulo)Highest
sa-east-1$9.6413
US East (N. Virginia)Lowest
us-east-1$5.6720
US East (Ohio)Lowest
us-east-2$5.6720
US West (Oregon)Lowest
us-west-2$5.6720