generated: '2026-08-26'
method: searched
source: https://www.hyperstack.cloud/gpu-pricing
docs: https://www.hyperstack.cloud/gpu-pricing
currency: USD
billing_model: usage-based
billing_granularity: per-minute
note: >-
  Hyperstack sells consumption, not subscription tiers - there is no free/pro/enterprise
  ladder. What it publishes instead is a full public rate card for three purchase modes
  (on-demand, reserved, spot) plus storage and networking, all in USD and billed per
  minute. Enterprise-scale contracts are quoted directly. The plans below model each
  purchase mode as a tier; the per-GPU rates are recorded under prices[]. Every figure is
  transcribed verbatim from the public pricing page on 2026-08-26 and rates change.
plan_count: 4
plans:
- name: On-Demand
  price_model: usage-based
  currency: USD
  billing_period: per-minute (rates quoted per hour)
  commitment: none
  quota: null
  overage: null
  limits:
  - 500 API requests per minute per source IP (see rate-limits/)
  prices:
  - resource: NVIDIA B300
    vram_gb: 288
    max_pcpus_per_gpu: 28
    max_ram_gb_per_gpu: 240
    price_per_hour: 7.40
  - resource: NVIDIA B200
    vram_gb: 192
    max_pcpus_per_gpu: 31
    max_ram_gb_per_gpu: 256
    price_per_hour: 6.00
  - resource: NVIDIA H200 SXM
    vram_gb: 141
    max_pcpus_per_gpu: 22
    max_ram_gb_per_gpu: 225
    price_per_hour: 3.99
  - resource: NVIDIA H100 SXM
    vram_gb: 80
    max_pcpus_per_gpu: 24
    max_ram_gb_per_gpu: 240
    price_per_hour: 3.20
  - resource: NVIDIA H100 NVLink
    vram_gb: 80
    price_per_hour: 2.60
  - resource: NVIDIA H100
    vram_gb: 80
    price_per_hour: 2.50
  - resource: NVIDIA RTX Pro 6000 SE
    vram_gb: 96
    price_per_hour: 1.85
  - resource: NVIDIA A100 SXM
    vram_gb: 80
    price_per_hour: 1.60
  - resource: NVIDIA A100 NVLink
    vram_gb: 80
    price_per_hour: 1.40
  - resource: NVIDIA A100
    vram_gb: 80
    price_per_hour: 1.35
  - resource: NVIDIA L40
    vram_gb: 48
    price_per_hour: 1.00
  - resource: NVIDIA A6000
    vram_gb: 48
    price_per_hour: 0.50
  - resource: NVIDIA A4000
    vram_gb: 16
    price_per_hour: 0.15
  - resource: n1-cpu-small (4 cores / 4 GB RAM / 100 GB root)
    price_per_hour: 0.35
  - resource: n1-cpu-medium (8 cores / 8 GB RAM / 100 GB root)
    price_per_hour: 0.65
  - resource: n1-cpu-large (16 cores / 16 GB RAM / 200 GB root)
    price_per_hour: 1.30
  - resource: n1-cpu-xlarge (16 cores / 32 GB RAM / 100 GB root)
    price_per_hour: 1.90
  - resource: n1-cpu-xxlarge (32 cores / 64 GB RAM / 100 GB root)
    price_per_hour: 3.74
- name: Reserved
  price_model: committed-term
  currency: USD
  billing_period: contract term
  commitment: long-term GPU reservation with discount; see https://docs.hyperstack.cloud/docs/billing/contracts
  prices:
  - resource: NVIDIA B200
    price_per_hour_from: 5.10
  - resource: NVIDIA H200 SXM
    price_per_hour_from: 2.79
  - resource: NVIDIA H100 SXM
    price_per_hour_from: 2.72
  - resource: NVIDIA H100 NVLink
    price_per_hour_from: 1.82
  - resource: NVIDIA H100
    price_per_hour_from: 1.75
  - resource: NVIDIA A100 SXM
    price_per_hour_from: 1.36
  - resource: NVIDIA RTX Pro 6000 SE
    price_per_hour_from: 1.30
  - resource: NVIDIA A100 NVLink
    price_per_hour_from: 0.98
  - resource: NVIDIA A100
    price_per_hour_from: 0.95
  - resource: NVIDIA L40
    price_per_hour_from: 0.70
  - resource: NVIDIA A6000
    price_per_hour_from: 0.35
  - resource: NVIDIA A4000
    price_per_hour_from: 0.11
- name: Spot VMs
  price_model: usage-based-interruptible
  currency: USD
  billing_period: per-minute (rates quoted per hour)
  commitment: none
  note: Low-cost interruptible capacity. https://docs.hyperstack.cloud/docs/virtual-machines/spot
  prices:
  - resource: NVIDIA H100 PCIe
    price_per_hour: 2.00
  - resource: NVIDIA RTX Pro 6000 SE
    price_per_hour: 1.48
  - resource: NVIDIA A100
    price_per_hour: 1.08
  - resource: NVIDIA L40
    price_per_hour: 0.80
  - resource: NVIDIA A6000
    price_per_hour: 0.40
- name: Enterprise / Secure Private Cloud
  price_model: contact-sales
  currency: USD
  price: null
  note: >-
    "For larger enterprise-scale contracts, please contact us directly." Single-tenant
    sovereign superclusters are quoted, not listed.
other_pricing:
- resource: Cloud-SSD volumes (SSVs)
  price: 0.000096774 per GB per hour (approximately $0.10 per TB per hour)
- resource: Snapshots
  price: Estimated at creation from VM size + attached volumes at current Cloud-SSD rates
  docs: https://docs.hyperstack.cloud/docs/billing/snapshots
- resource: Public IP address
  price: 0.00672043 per hour
- resource: Egress / ingress traffic
  price: free
- resource: On-Demand Kubernetes master node
  price: free
machine_readable_pricing:
  available: true
  note: >-
    Pricing is also queryable through the API rather than only the marketing page - the
    Pricebook (GET /pricebook) and Calculate endpoints return resource rates by flavor
    and region. This is unusual and materially useful to an agent budgeting a run.
  operations:
  - Get_pricebook
  - Calculate_resource_billing_rate
  docs: https://docs.hyperstack.cloud/docs/api-reference/pricebook