Cloud GPU Prices

Data interface · Agent and API manual

Cloud GPU Pricing Catalog

Filter accepted current offerings without treating incomplete component prices as totals. Every result retains its source, verification time, geography state, and comparison eligibility.

More filters
Reset

Hourly price sorting intentionally requires “complete comparable totals” and exactly one currency. Repeat parameters such as provider are supported by the API and URL contract.

Results

15 matching offerings · 28 commercial variants · filtered

JSON Markdown
Beam.cloud
H200
Serverless
NVIDIA H200 · fixed · 1 GPU · 141 GB VRAM
  • USD 0.000013/core second
    cpu core time
  • USD 0.001136/second
    gpu time
  • USD 0.000002/gib second
    memory time
Not a complete total
Standard
scale to zero · standard · unspecified · additive components
additive price not total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
unpriced custom plan: The Serverless committed-spend option is presented as “With committed spend Book a call →” without a published amount; it was omitted.
Cerebrium
H200
Serverless
NVIDIA H200 · fixed · 1 GPU · VRAM unknown
  • USD 0.000007/core second
    cpu core time
  • USD 0.001166/second
    gpu time
  • USD 0.000002/gb second
    memory time
Not a complete total
Standard
scale to zero · standard · unspecified · additive components
additive price not total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
Fireworks AI
H200 141 GB GPU
On-Demand Deployments
NVIDIA H200 · fixed · 1 GPU · 141 GB VRAM
  • USD 7/hour
    gpu time
Comparable total
Always on
always on · standard · dedicated · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
Fireworks AI
H200 141 GB GPU
On-Demand Deployments
NVIDIA H200 · fixed · 1 GPU · 141 GB VRAM
  • USD 7/hour
    gpu time
Comparable total
Scale to zero
scale to zero · standard · dedicated · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
Hugging Face
AWS NVIDIA H200 x2
Inference Endpoints
NVIDIA H200 · fixed · 2 GPU · 282 GB VRAM
  • USD 10/hour
    compute time
Comparable total
Always on
always on · standard · dedicated · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
ambiguous hardware: GPU rows do not publish system memory, so systemMemoryGb remains null.
Hugging Face
AWS NVIDIA H200 x2
Inference Endpoints
NVIDIA H200 · fixed · 2 GPU · 282 GB VRAM
  • USD 10/hour
    compute time
Comparable total
Scale to zero
scale to zero · standard · dedicated · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
ambiguous hardware: GPU rows do not publish system memory, so systemMemoryGb remains null.
Hugging Face
AWS NVIDIA H200 x8
Inference Endpoints
NVIDIA H200 · fixed · 8 GPU · 1128 GB VRAM
  • USD 40/hour
    compute time
Comparable total
Always on
always on · standard · dedicated · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
ambiguous hardware: GPU rows do not publish system memory, so systemMemoryGb remains null.
Hugging Face
AWS NVIDIA H200 x8
Inference Endpoints
NVIDIA H200 · fixed · 8 GPU · 1128 GB VRAM
  • USD 40/hour
    compute time
Comparable total
Scale to zero
scale to zero · standard · dedicated · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
ambiguous hardware: GPU rows do not publish system memory, so systemMemoryGb remains null.
Hugging Face
AWS NVIDIA H200 x4
Inference Endpoints
NVIDIA H200 · fixed · 4 GPU · 564 GB VRAM
  • USD 20/hour
    compute time
Comparable total
Always on
always on · standard · dedicated · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
ambiguous hardware: GPU rows do not publish system memory, so systemMemoryGb remains null.
Hugging Face
AWS NVIDIA H200 x4
Inference Endpoints
NVIDIA H200 · fixed · 4 GPU · 564 GB VRAM
  • USD 20/hour
    compute time
Comparable total
Scale to zero
scale to zero · standard · dedicated · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
ambiguous hardware: GPU rows do not publish system memory, so systemMemoryGb remains null.
Hugging Face
AWS NVIDIA H200 x1
Inference Endpoints
NVIDIA H200 · fixed · 1 GPU · 141 GB VRAM
  • USD 5/hour
    compute time
Comparable total
Always on
always on · standard · dedicated · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
ambiguous hardware: GPU rows do not publish system memory, so systemMemoryGb remains null.
Hugging Face
AWS NVIDIA H200 x1
Inference Endpoints
NVIDIA H200 · fixed · 1 GPU · 141 GB VRAM
  • USD 5/hour
    compute time
Comparable total
Scale to zero
scale to zero · standard · dedicated · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
ambiguous hardware: GPU rows do not publish system memory, so systemMemoryGb remains null.
Koyeb
2x H200
Serverless Compute
NVIDIA H200 · fixed · 2 GPU · 282 GB VRAM
  • USD 6/hour
    compute time
Comparable total
Always-on
always on · standard · unspecified · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
unpriced custom plan: Long-term reservations have no public exact price and were omitted.
Koyeb
2x H200
Serverless Compute
NVIDIA H200 · fixed · 2 GPU · 282 GB VRAM
  • USD 6/hour
    compute time
Comparable total
Scale-to-Zero
scale to zero · standard · unspecified · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
unpriced custom plan: Long-term reservations have no public exact price and were omitted.
Koyeb
8x H200
Serverless Compute
NVIDIA H200 · fixed · 8 GPU · 1128 GB VRAM
  • USD 24/hour
    compute time
Comparable total
Always-on
always on · standard · unspecified · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
unpriced custom plan: Long-term reservations have no public exact price and were omitted.
Koyeb
8x H200
Serverless Compute
NVIDIA H200 · fixed · 8 GPU · 1128 GB VRAM
  • USD 24/hour
    compute time
Comparable total
Scale-to-Zero
scale to zero · standard · unspecified · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
unpriced custom plan: Long-term reservations have no public exact price and were omitted.
Koyeb
H200
Serverless Compute
NVIDIA H200 · fixed · 1 GPU · 141 GB VRAM
  • USD 3/hour
    compute time
Comparable total
Always-on
always on · standard · unspecified · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
unpriced custom plan: Long-term reservations have no public exact price and were omitted.
Koyeb
H200
Serverless Compute
NVIDIA H200 · fixed · 1 GPU · 141 GB VRAM
  • USD 3/hour
    compute time
Comparable total
Scale-to-Zero
scale to zero · standard · unspecified · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
unpriced custom plan: Long-term reservations have no public exact price and were omitted.
Koyeb
4x H200
Serverless Compute
NVIDIA H200 · fixed · 4 GPU · 564 GB VRAM
  • USD 12/hour
    compute time
Comparable total
Always-on
always on · standard · unspecified · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
unpriced custom plan: Long-term reservations have no public exact price and were omitted.
Koyeb
4x H200
Serverless Compute
NVIDIA H200 · fixed · 4 GPU · 564 GB VRAM
  • USD 12/hour
    compute time
Comparable total
Scale-to-Zero
scale to zero · standard · unspecified · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
unpriced custom plan: Long-term reservations have no public exact price and were omitted.
Modal
Nvidia H200
Serverless
NVIDIA H200 · fixed · 1 GPU · VRAM unknown
  • USD 0.000013/core second → USD 0.00002/core second effective
    cpu core time
  • USD 0.001261/second → USD 0.001892/second effective
    gpu time
  • USD 0.000002/gib second → USD 0.000003/gib second effective
    memory time
Not a complete total
Broad region
scale to zero · standard · unspecified · additive components
additive price not total
EU compute: yes EU-constrained: yes EU-only: no
current Verified 17 Jul 2026 Official price source
5 evidence-backed claims
Modal
Nvidia H200
Serverless
NVIDIA H200 · fixed · 1 GPU · VRAM unknown
  • USD 0.000013/core second → USD 0.000023/core second effective
    cpu core time
  • USD 0.001261/second → USD 0.002207/second effective
    gpu time
  • USD 0.000002/gib second → USD 0.000004/gib second effective
    memory time
Not a complete total
Narrow region
scale to zero · standard · unspecified · additive components
additive price not total
EU compute: yes EU-constrained: yes EU-only: no
current Verified 17 Jul 2026 Official price source
5 evidence-backed claims
Modal
Nvidia H200
Serverless
NVIDIA H200 · fixed · 1 GPU · VRAM unknown
  • USD 0.000013/core second
    cpu core time
  • USD 0.001261/second
    gpu time
  • USD 0.000002/gib second
    memory time
Not a complete total
Standard region
scale to zero · standard · unspecified · additive components
additive price not total
EU compute: yes EU-constrained: yes EU-only: no
current Verified 17 Jul 2026 Official price source
5 evidence-backed claims
OVHcloud
h200-4-gpu
AI Deploy
NVIDIA H200 · fixed · 4 GPU · 525 GB VRAM
  • USD 27.28/hour
    compute time
Comparable total
h200-4-gpu always-on
always on · standard · dedicated · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
OVHcloud
h200-4-gpu
AI Deploy
NVIDIA H200 · fixed · 4 GPU · 525 GB VRAM
  • USD 27.28/hour
    compute time
Comparable total
h200-4-gpu scale-to-zero
scale to zero · standard · dedicated · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
Runpod
H200
Serverless
NVIDIA H200 · fixed · 1 GPU · 140 GB VRAM
  • USD 5.93/hour
    gpu time
Comparable total
Flex
scale to zero · standard · unspecified · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
unpriced custom plan: Active worker discounts are available through sales inquiry and are omitted because no public price is published.
Verda
1× H200 SXM5
GPU Instances
NVIDIA H200 · fixed · 1 GPU · 141 GB VRAM
  • USD 4/hour
    compute time
Comparable total
On-demand
instance rental · standard · dedicated · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source
Verda
1× H200 SXM5
GPU Instances
NVIDIA H200 · fixed · 1 GPU · 141 GB VRAM
  • USD 1.4/hour
    compute time
Comparable total
Spot
instance rental · spot · dedicated · inclusive total
EU compute: unknown EU-constrained: unknown EU-only: unknown
current Verified 17 Jul 2026 Official price source