GPUBank vs CloudRift
CloudRift Inc. is a Delaware company that sells GPU orchestration software to datacenter operators and runs a public GPU rental marketplace on partner datacenters in North America, Europe and Asia. It publishes the H100 80GB only as reserved capacity, at $1.70 per GPU-hour for 1 month and $1.33 for 3 months, with no on-demand price; GPUBank rents an H100 SXM on demand at $1.85. Both bill per second with no egress fees; CloudRift is cheaper on long H100 and H200 reservations, while GPUBank sells those GPUs on demand and takes crypto.
GPUBank
Tallinn, Estonia
H200 SXM
$2.49 / h
- vCPU, RAM, NVMe, IPv4 and traffic included
- Per-second billing after the health check
- Top up in Bitcoin, USDT and 12+ other coins
CloudRift
Santa Clara, United States
H200 SXM · reserved
$2.38 / h
- H100 at $1.33 and H200 at $2.38 per GPU-hour on 3-month reservations
- V100 from $0.25 per GPU-hour and AMD MI350X in the catalogue
- Pay-per-token LLM inference API on the same account
CloudRift
Marketplace
26 points compared26 of their pages readprices last checked on 29 September 2026page updated 2 November 2026
The verdict
Long reservations and old GPUs at CloudRift. On-demand H100 and H200, and crypto top-ups, at GPUBank.
CloudRift is cheaper for teams that can prepay 3 months: its H100 reservation is $1.33 and its H200 is $2.38 per GPU-hour. It also lists V100 from $0.25, MI350X and a pay-per-token inference API. For on-demand work, GPUBank costs 10 to 41 % less on the five GPUs both sell on demand, from the A100 SXM to the RTX 4090, and CloudRift shows H100, H200, L40S and RTX PRO 6000 as Upon request on its pricing table. At check time its public API showed stock only for V100 types, and Stop terminates a rental, where GPUBank keeps the disk.
01The real cost
The same GPUs, side by side, per hour.
Every GPU model both of us rent, at each provider's own published price for one card, in US dollars, excluding VAT. Prices last checked on 29 September 2026.
| GPU | CloudRift per hour | GPUBank per hour | Difference | 100 h at CloudRift | 100 h at GPUBank |
|---|---|---|---|---|---|
| $2.38 | $2.49 | +5 % | $238.00 | $249.00 | |
| $1.10 | $1.15 | +5 % | $110.00 | $115.00 | |
| $1.05 | $0.95 | −10 % | $105.00 | $95.00 | |
| $0.54 | $0.55 | +2 % | $54.00 | $55.00 | |
| $0.60 | $0.48 | −20 % | $60.00 | $48.00 | |
| $0.39 | $0.23 | −41 % | $39.00 | $23.00 |
CloudRift publishes no on-demand price for H200, RTX PRO 6000 or L40S on its pricing table: those rows are 3-month reserved rates, paid upfront. Its API lists on-demand rates for the last two, with 0 nodes available.
Where CloudRift publishes no on-demand price for a model, its lowest published reserved price is used, as labelled on the row.
An inference box, a month, 720 h, on an L40S
CloudRift
$388.80
GPUBank
$396.00
Costs $7.20 more at GPUBank (+2 %)
Image generation, 200 h on an RTX 4090
CloudRift
$78.00
GPUBank
$46.00
Saves $32.00 at GPUBank (−41 %)
A long-context job, 40 h on an H200
CloudRift
$95.20
GPUBank
$99.60
Costs $4.40 more at GPUBank (+5 %)
02Their catalogue
Every GPU CloudRift lists.
CloudRift lists 11 GPU models. The 9 below are the ones recorded in our check, each at its lowest public price per GPU-hour. 6 of them are also rented at GPUBank (in blue).
H200141 GB · reserved$2.38- H10080 GB · reserved$1.33
A100 SXM 80GB80 GB · on demand$1.05
L40S48 GB · reserved$0.54
V10016 GB · on demand$0.25
RTX PRO 600096 GB · reserved$1.10
RTX 509032 GB · on demand$0.60
RTX 409024 GB · on demand$0.39
MI350X288 GB · on demand$4.00At GPUBank
vCPU, RAM, NVMe and traffic included. All prices
03What the price includes
What appears on the bill besides the GPU.
The same H200 SXM session of 2 h 14 m 03 s, line by line, as each provider describes its billing.
CloudRift
2 h 14 m 03 s on H200 SXM
- GPU · $2.38 / h
- $5.32
- vCPU and RAM
- Included
- Storage
- Not published
- Egress
- Included
- Public IPv4
- Not published
- Billing granularity
- Per second, from VM readiness
- Minimum charge
- None
GPUBank
2 h 14 m 03 s on H200 SXM
- GPU · $2.49 / h
- $5.56
- vCPU and RAM
- Included
- Storage
- Local NVMe included; persistent disk $0.03 per GB-month
- Egress
- Included
- Public IPv4
- Included
- Billing granularity
- Per second, from the health check
- Minimum charge
- None
While stopped
What keeps running on the bill.
CloudRift
Stop terminates the rental, local data lost.
An API pause keeps resources and billing running.
GPUBank
Only the persistent disk, $0.03 per GB-month. Compute, IPv4 and scratch cost nothing.
04Where the GPUs run
8 countries against 4, and where they are.
Countries where CloudRift says its GPUs run, from its own region list or documentation, against GPUBank's four regions. A pin marks a country, not the exact site.
CloudRift8
United States
Canada
United Kingdom
Italy
France
Taiwan
South Korea
Kazakhstan
GPUBank4
Paris
Frankfurt
Ashburn
Singapore
CloudRift · 8 countriesGPUBank · 4 regions
05Paying and signing up
How each one takes your money.
CloudRift
Payments and sign-up
Payment methods
Minimum top-up Not published
To start
- 1Email and password
Billing modelPrepaid credits
GPUBank
Payments and sign-up
Payment methods
A prepaid USD balance, topped up in crypto. Every GPU-second is drawn from it.
Minimum top-up $25
To start
- 1Email and password
- 2Top up in crypto
- 3Launch a GPU
Billing modelPrepaid balance
06Trust and support
Certificates, uptime and who answers.
| GPUBank | ||
|---|---|---|
| SOC 2 | Not published | Not published |
| ISO 27001 | No | Not published |
| HIPAA | Not published | No |
| GDPR DPA | Not published | Published DPA, applies automatically |
| Availability target | None; 99.99 % headline without terms | 99.9 % monthly, per region |
| Status page | status.cloudrift.ai | Public status page |
| Support | Discord, email, contact form, phone | Tickets from a $25 balance |
07Their site, our site
What you see before you pay.
Both sites as a visitor sees them, without an account: the CloudRift pricing page captured on 29 September 2026, and GPUBank's live board. Cookie banners and chat bubbles are closed; nothing else is edited.

- 1The pricing table shows 11 GPU models with on-demand, 1-month and 3-month rates, visible without an account.
- 2L40S, RTX PRO 6000, RTX PRO 6000 Max-Q, H100 and H200 show Upon request in the on-demand column.
- 3The calculator lists H100 at $1.70 and H200 at $2.50, their 1-month reserved rates, next to on-demand rates of other models.

- 1Live stock per region and every price, without an account.
- 2Per-hour and per-second prices, with vCPU, RAM, NVMe, IPv4 and traffic in them.
- 3A machine-readable price feed at /prices.json.
08The spec sheet
26 points, read off both sites.
GPUBank aheadCloudRift aheadEven
| Criterion | GPUBank | |
|---|---|---|
| Price and catalogue | ||
| H100 SXM, per GPU-hourNo on-demand H100 price. GPUBank: $1.85 on demand, $1.48 reserved for 1 month. | $1.85, all-in | $1.70 (1 month), $1.33 (3 months), reserved only |
| Cheapest GPU-hour | $0.23, RTX 4090 | $0.25, V100 16GB |
| GPU models listed | 10 | 11Ahead |
| What the GPU price includes | vCPU, RAM, NVMe, IPv4, traffic | Compute, local storage, networking, support |
| Billing | ||
| Billing granularity | Per second | Per second, from VM readiness |
| Minimum charge | None | None |
| Stopped instanceAn API pause keeps resources and billing running. | Disk only | Stop terminates the rental, local data lost |
| Persistent storagePersistent volumes exist in some datacenters and are deleted when the balance reaches zero. | $0.03 per GB-month | Not published |
| Egress traffic | Included | Included |
| Public IPv4 | Included | Not published |
| Billing model | Prepaid balance | Prepaid credits |
| Payment and sign-up | ||
| Crypto payments | Bitcoin, USDT, Ethereum and 12+ more | Not accepted |
| Card payments | No | Visa, Mastercard, Amex, CashApp, bank transferAhead |
| Minimum depositStripe deposits are capped at 2,500 USD; larger amounts go by invoice. | $25 | Not published |
| Platform | ||
| API and CLI | REST API, CLI, Python SDK | REST API, rift CLI, Python package, MCP |
| Terraform | Official provider | No provider found |
| Serverless GPUs | No | Pay-per-token LLM inference APIAhead |
| Managed Kubernetes | No | No |
| Ready-made templates | 25 | PyTorch, vLLM, ComfyUI images, custom recipes |
| Interruptible or spot | Yes, 40 % off | No |
| GPUs per instance | Up to 8 | Up to 8; 10 on RTX 4090, 16 on V100Ahead |
| Multi-node interconnect | Not offered | NVLink on A100 and V100; no InfiniBand |
| Network and regions | ||
| Regions | 4 | 8 locations on partner datacentersAhead |
| Countries | France, Germany, US, Singapore | 8, incl. US, Canada, UK, Italy, Taiwan, South KoreaAhead |
| EU regionsGPUBank has two EU regions, Paris and Frankfurt. | Paris, Frankfurt | Italy; a French site is not on the marketplace |
| Trust and support | ||
| Availability target | 99.9 % monthly | None; 99.99 % headline without terms |
| Compliance reportsThe home page shows a SOC 2 Certified badge while the security page says In Progress. | None published | SOC 2 Type II in progress |
| Public status page | Yes | status.cloudrift.ai |
| Support | Tickets from a $25 balance | Discord, email, contact form, phoneAhead |
09The right fit
Different needs. A clear choice.
What CloudRift does better
- H100 at $1.33 and H200 at $2.38 per GPU-hour on 3-month reservations
- V100 from $0.25 per GPU-hour and AMD MI350X in the catalogue
- Pay-per-token LLM inference API on the same account
- Partner datacenters in 8 countries, including Taiwan and South Korea
- Cards, CashApp and bank transfer
Choose GPUBank if
- H100 SXM on demand at $1.85; CloudRift lists H100 as Upon request
- A100 SXM at $0.95 against $1.05, RTX 5090 at $0.48 against $0.60
- Stop an instance and keep its disk; CloudRift Stop ends the rental
- Interruptible at 40 % off; CloudRift has no spot tier
- Crypto top-ups and a Terraform provider
Moving from CloudRift takes an afternoon.
CloudRift instances are KVM VMs or Docker containers with SSH access and no egress fees, so data leaves at no cost on its side. Stop ends a rental and wipes local storage, so copy data out first.
- 1Launch the same GPU on GPUBank with a persistent disk in the region you need.
- 2Copy data from the running CloudRift instance with rsync over SSH, including attached volumes.
- 3Run the same Docker image or a matching template, then stop the CloudRift rental.
# 1. a GPU and a disk on GPUBank gpubank launch --gpu h200-sxm --region eu-paris --disk data # 2. copy your data from the old machine rsync -azP old-host:/workspace/ root@<new-ip>:/workspace/ # 3. run the same image or template ssh root@<new-ip> 'cd /workspace && ./start.sh'
A few things you might be wondering.
It depends on the term. CloudRift publishes no on-demand H100 price; it reserves H100 80GB at $1.70 per GPU-hour for 1 month and $1.33 for 3 months, paid upfront. GPUBank rents an H100 SXM at $1.85 on demand, or $1.48 reserved for 1 month.
Yes. CloudRift bills per second while running, with no minimum, and VM billing starts when the VM signals readiness. GPUBank also bills per second and starts the meter after the health check.
Stop terminates the rental and the ephemeral storage is lost. Persistent volumes keep billing, at a price that is not published. On GPUBank a stopped instance keeps its persistent disk at $0.03 per GB-month.
No. Its terms list Visa, Mastercard, American Express, CashApp and bank transfer, and Stripe deposits are capped at 2,500 USD. GPUBank is paid from a USD balance topped up in Bitcoin, USDT, Ethereum and more than a dozen other coins.
At check time, CloudRift's public API showed available nodes only for V100 instance types. Capacity sits in third-party partner datacenters, and the pricing page notes that availability varies by demand and region.
Without. Every price on this page excludes VAT and sales tax, for CloudRift and for GPUBank, so the two columns compare like for like. Tax is added at checkout according to where the buyer is based.
Sources
Every figure about CloudRift was read on its own pages or official documentation. Prices last checked on 29 September 2026.
- 01cloudrift.ai/pricingGPU price table, calculator, reservation discounts, billing FAQ, egress
- 02cloudrift.ai/site observations, SOC 2 claim, GPU cards
- 03cloudrift.ai/gpu-rentals99.99% uptime headline, per-second billing, 8 locations, VM/container, Tier 3 facilities, price FAQ
- 04cloudrift.ai/rtx5090alternative RTX 5090 price quotes
- 05cloudrift.ai/mi350xalternative MI350X price quotes
- 06cloudrift.ai/v100V100 prices, NVLink 2.0
- 07cloudrift.ai/a100A100 per-GPU resources, NVLink 3.0
- 08cloudrift.ai/aboutfounders, deployment countries
- 09cloudrift.ai/contactlegal entity, address, contact channels
- 10cloudrift.ai/securitySOC 2 Type II in progress, Vanta
- 11cloudrift.ai/termspayment methods, no refunds, no availability guarantee
- 12cloudrift.ai/ai-grantfree credits 100 to 1, 000 USD
- 13cloudrift.ai/for-enterprisebare metal, MIG, platform scope
- 14docs.cloudrift.ai/features/volumesvolume behaviour, deletion at zero balance
- 15docs.cloudrift.ai/features/reservationsreservation conversion, Stripe 2, 500 USD limit
- 16docs.cloudrift.ai/troubleshooting/volumes-and-reservationsstop = terminate, volume billing after termination
- 17docs.cloudrift.ai/troubleshooting/setup-and-generalbilling stops at Stop, Discord support, Stripe billing
- 18docs.cloudrift.ai/extras/rest-apiAPI capabilities, MCP endpoint
What we could not verify (8)
- Persistent volume price per GB-month not published
- Public IPv4 pricing not published
- Minimum deposit not published (Stripe upper limit of 2,500 USD documented)
- H100 form factor (SXM or PCIe) not published; on-demand rates for L40S, RTX PRO 6000, H100 and H200 not on the pricing page (L40S and RTX PRO 6000 read from the public API only)
- Unit of cost_per_hour in the public API is not documented; US cents inferred by matching pricing-table values
- Founding year and funding not published on the pages read
- No SLA document; the 99.99% figure has no stated terms
- No Terraform provider or Kubernetes offering found in the docs
In short
GPUBank vs CloudRift, in facts.
The essentials of this comparison in one table, for quick readers and AI assistants. Prices last checked on 29 September 2026.
- H100 SXM, per GPU-hour
- CloudRift: $1.70 (1 month), $1.33 (3 months), reserved only
GPUBank: $1.85, all-in - Cheapest GPU-hour
- CloudRift: $0.25, V100 16GB
GPUBank: $0.23, RTX 4090 - Billing granularity
- CloudRift: Per second, from VM readiness
GPUBank: Per second - Stopped instance
- CloudRift: Stop terminates the rental, local data lost
GPUBank: Disk only - Persistent storage
- CloudRift: Not published
GPUBank: $0.03 per GB-month - Egress traffic
- CloudRift: Included
GPUBank: Included - Crypto payments
- CloudRift: Not accepted
GPUBank: Bitcoin, USDT, Ethereum and 12+ more - Card payments
- CloudRift: Visa, Mastercard, Amex, CashApp, bank transfer
GPUBank: No - Minimum deposit
- CloudRift: Not published
GPUBank: $25 - GPU models listed
- CloudRift: 11
GPUBank: 10 - Regions
- CloudRift: 8 locations on partner datacenters
GPUBank: 4 - Serverless GPUs
- CloudRift: Pay-per-token LLM inference API
GPUBank: No - Availability target
- CloudRift: None; 99.99 % headline without terms
GPUBank: 99.9 % monthly - Compliance reports
- CloudRift: SOC 2 Type II in progress
GPUBank: None published - Support
- CloudRift: Discord, email, contact form, phone
GPUBank: Tickets from a $25 balance - Headquarters
- CloudRift: Santa Clara, United States
GPUBank: Tallinn, Estonia
How this page was made
Every price and policy of CloudRift was read on its own website, documentation or public API on 29 September 2026, with the address of each page listed above. A value we could not read on a primary source is marked "Not published" rather than estimated. GPUBank figures come from this site and its terms. Prices change: check both providers before you buy. CloudRift, its logo and its website belong to their owner and are shown only to identify the service compared. GPUBank is not affiliated with CloudRift.