last_verified: '2026-04-08'
source: https://apexapps.oracle.com/pls/apex/cetools/api/v1/products/
description: OCI Compute pricing for estimation purposes. Auto-refreshed from the
  Oracle public pricing API by tools/refresh_sku_catalog.py --refresh-domain compute.
  GPU shapes, secure desktops, estimation_helpers, and discounts are NOT auto-refreshed.
currency: USD
---
flexible_shapes:
  VM.Standard.E6.Flex:
    ocpu_per_hour: 0.03
    memory_per_gb_hour: 0.002
    monthly_730h:
      ocpu: 21.9
      memory_per_gb: 1.46
    notes: AMD EPYC (latest gen), best price/performance for general workloads
  VM.Standard.E5.Flex:
    ocpu_per_hour: 0.03
    memory_per_gb_hour: 0.002
    monthly_730h:
      ocpu: 21.9
      memory_per_gb: 1.46
    notes: AMD EPYC, general-purpose workloads
  VM.Standard.E4.Flex:
    ocpu_per_hour: 0.025
    memory_per_gb_hour: 0.0015
    monthly_730h:
      ocpu: 18.25
      memory_per_gb: 1.09
    notes: AMD EPYC (previous gen)
  VM.Standard.E3.Flex:
    ocpu_per_hour: 0.025
    memory_per_gb_hour: 0.0015
    monthly_730h:
      ocpu: 18.25
      memory_per_gb: 1.09
    notes: AMD EPYC (older gen)
  VM.Standard3.Flex:
    ocpu_per_hour: 0.0336
    memory_per_gb_hour: 0.0022
    monthly_730h:
      ocpu: 24.53
      memory_per_gb: 1.61
    notes: Intel Xeon, for Intel-optimized workloads
  VM.Standard.A2.Flex:
    ocpu_per_hour: 0.014
    memory_per_gb_hour: 0.002
    monthly_730h:
      ocpu: 10.22
      memory_per_gb: 1.46
    notes: Ampere Arm (latest A2), best cost for Arm-compatible workloads
  VM.Standard.A1.Flex:
    ocpu_per_hour: 0.0127
    memory_per_gb_hour: 0.0008
    monthly_730h:
      ocpu: 9.27
      memory_per_gb: 0.58
    free_tier: 3,000 OCPU-hours + 18,000 GB-hours per month
    notes: Ampere Arm A1, eligible for Always Free tier
  VM.Standard.x86.Generic:
    ocpu_per_hour: 0.0252
    memory_per_gb_hour: 0.0017
    notes: Generic x86 pricing when specific shape not yet selected
  VM.Optimized3.Flex:
    ocpu_per_hour: 0.054
    memory_per_gb_hour: 0.0015
    monthly_730h:
      ocpu: 39.42
      memory_per_gb: 1.09
    notes: Highest single-thread performance
bare_metal_shapes:
  BM.Standard.E5:
    ocpu_per_hour: 0.03
    notes: AMD EPYC bare metal
  BM.Standard.E4:
    ocpu_per_hour: 0.025
    notes: AMD EPYC bare metal (previous gen)
  BM.Standard.E3:
    ocpu_per_hour: 0.025
    notes: AMD EPYC bare metal (older gen)
  BM.Standard3:
    ocpu_per_hour: 0.0504
    notes: Intel Xeon bare metal
  BM.Standard.A1:
    ocpu_per_hour: 0.0127
    notes: Ampere Arm bare metal
  BM.Standard.x9:
    ocpu_per_hour: 0.04
    notes: Intel Xeon x9 bare metal
gpu_shapes:
  BM.GPU.B300:
    per_gpu_hour: 26.88
    gpu_count: 8
    gpu_model: NVIDIA B300
    notes: Blackwell latest, HPC/AI training
  BM.GPU.B200:
    per_gpu_hour: 20.16
    gpu_count: 8
    gpu_model: NVIDIA B200
    notes: Blackwell, large-scale training
  BM.GPU.H200:
    per_gpu_hour: 18.54
    gpu_count: 8
    gpu_model: NVIDIA H200
    notes: Hopper H200, large model training
  BM.GPU.H100:
    per_gpu_hour: 12.96
    gpu_count: 8
    gpu_model: NVIDIA H100
    notes: Hopper H100, training workloads
  BM.GPU.A100:
    per_gpu_hour: 6.72
    gpu_count: 8
    gpu_model: NVIDIA A100 40GB/80GB
    notes: Training and inference, 40GB or 80GB HBM2e
  VM.GPU.A10.1:
    per_gpu_hour: 0.72
    gpu_count: 1
    gpu_model: NVIDIA A10
    notes: Inference workloads, cost-effective
  VM.GPU.A10.2:
    per_gpu_hour: 0.72
    gpu_count: 2
    gpu_model: NVIDIA A10
    notes: Inference workloads
  BM.GPU.L40S:
    per_gpu_hour: 2.42
    gpu_count: 4
    gpu_model: NVIDIA L40S
    notes: Ada Lovelace, inference and graphics
secure_desktops:
  per_desktop_month: 1750.0
  notes: Managed virtual desktop infrastructure
estimation_helpers:
  typical_app_server:
    config: VM.Standard.E5.Flex, 4 OCPU, 64 GB
    monthly: 181.04
  typical_web_server:
    config: VM.Standard.E5.Flex, 2 OCPU, 16 GB
    monthly: 67.16
  typical_bastion:
    config: VM.Standard.E5.Flex, 1 OCPU, 8 GB
    monthly: 33.58
  typical_arm_app_server:
    config: VM.Standard.A1.Flex, 4 OCPU, 24 GB
    monthly: 51.1
discounts:
  preemptible: ~50% off on-demand
  capacity_reservation: ~15% off on-demand (85% of list)
  reserved_1yr: ~35% off on-demand
  reserved_3yr: ~55% off on-demand
  universal_credits: Volume discounts on committed spend
os_pricing:
  free:
  - Oracle Linux
  - CentOS
  - Ubuntu
  - Oracle Autonomous Linux
  paid:
    windows_server:
      notes: Pricing varies by edition, added to compute cost