GPUBank vs Massed Compute
Massed Compute is a US GPU cloud, established in 2021, that owns its hardware in Tier III data centers and rents NVIDIA VMs of 1 to 8 GPUs, from A30 to B300, each with a ThinLinc remote desktop. Its H100 SXM5 costs $2.89 per GPU-hour on one GPU and $3.14 on an 8-GPU VM; GPUBank charges $1.85. Massed Compute has a wider catalogue, card payments and SOC 2 claims; GPUBank can stop an instance and keep its disk, runs in four countries and bills by the second.
GPUBank
Tallinn, Estonia
H100 SXM
$1.85 / h
- vCPU, RAM, NVMe, IPv4 and traffic included
- Per-second billing after the health check
- Top up in Bitcoin, USDT and 12+ other coins
vCPU, RAM, NVMe, IPv4 and traffic included
Massed Compute
Las Vegas, United States · since 2021
H100 SXM
$2.89 / h
- 19 GPU models, including A30, RTX A5000, L40 and RTX 6000 Ada
- ThinLinc remote desktop included on every VM
- Cards through Stripe, with PayPal listed in the Terms
Massed Compute
GPU cloud
United States28 points compared39 of their pages readprices last checked on 29 September 2026page updated 2 November 2026
The verdict
A wide catalogue and a desktop on every VM at Massed Compute. Lower prices and persistent disks at GPUBank.
Massed Compute suits US users who want older or workstation cards such as the A30, RTX A5000 or RTX 6000 Ada, a browser desktop on each VM, card payments, and quoted bare metal or InfiniBand clusters. For the seven GPUs both of us list, GPUBank costs 24 to 51 % less per GPU-hour. Massed Compute has no stop action and no volumes, so terminating a VM deletes all its data; GPUBank stops an instance and bills only its disk. Massed Compute takes Bitcoin only, without auto-recharge; GPUBank takes more than a dozen coins.
Massed ComputeGPUBank 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 | Massed Compute per hour | GPUBank per hour | Difference | 100 h at Massed Compute | 100 h at GPUBank |
|---|---|---|---|---|---|
 B200 SXM180 GB |
$5.43 | $4.15 | −24 % | $543.00 | $415.00 |
 H200 SXM141 GB |
$3.62 | $2.49 | −31 % | $362.00 | $249.00 |
 H100 SXM80 GB |
$2.89 | $1.85 | −36 % | $289.00 | $185.00 |
 H100 PCIe80 GB |
$2.73 | $1.35 | −51 % | $273.00 | $135.00 |
 RTX PRO 600096 GB |
$2.19 | $1.15 | −47 % | $219.00 | $115.00 |
 A100 SXM80 GB |
$1.38 | $0.95 | −31 % | $138.00 | $95.00 |
 L40S48 GB |
$0.97 | $0.55 | −43 % | $97.00 | $55.00 |
Their H200 is the H200 NVL (PCIe form factor), shown against our H200 SXM. Their H100 (80GB) has no stated form factor; we treat it as PCIe. The B200 figure is an 8-GPU VM price divided by 8.
A fine-tune, 40 h on an H100 SXM
Massed Compute
$115.60
GPUBank
$74.00
Saves $41.60 at GPUBank (−36 %)
An inference box, a month, 720 h, on an L40S
Massed Compute
$698.40
GPUBank
$396.00
Saves $302.40 at GPUBank (−43 %)
A long-context job, 40 h on an H200
Massed Compute
$144.80
GPUBank
$99.60
Saves $45.20 at GPUBank (−31 %)
02Their catalogue
Every GPU Massed Compute lists.
Massed Compute lists 19 GPU models. The 16 below are the ones recorded in our check, each at its lowest public price per GPU-hour. 7 of them are also rented at GPUBank (in blue).
- Blackwell and Hopper6
- B300on demand$6.60
- B200on demand$5.43
- H200141 GB · on demand$3.62
- H100 SXM80 GB · on demand$2.89
- H100 NVL94 GB · on demand$2.92
- H100 PCIe80 GB · on demand$2.73
- Ampere, Ada and older data-centre5
- A100 SXM 80GB80 GB · on demand$1.38
- A10080 GB · on demand · variant not stated$1.35
- L40S48 GB · on demand$0.97
- L4048 GB · on demand$0.86
A3024 GB · on demand$0.35
- Workstation5
- RTX PRO 600096 GB · on demand$2.19
RTX 6000 Ada48 GB · on demand$0.79
RTX A600048 GB · on demand$0.55
RTX A500024 GB · on demand$0.44- RTX PRO 4500 Blackwell32 GB · on demand$0.92
03What the price includes
What appears on the bill besides the GPU.
The same H100 SXM session of 2 h 14 m 03 s, line by line, as each provider describes its billing.
Massed Compute
2 h 14 m 03 s on H100 SXM
- GPU · $2.89 / h
- $6.46
- vCPU and RAM
- Included
- Storage
- Local disk bundled; no volume product
- Egress
- Included
- Public IPv4
- Not published
- Billing granularity
- Per minute, per the docs
- Minimum charge
- 1 hour on marketing pages; none in docs
GPUBank
2 h 14 m 03 s on H100 SXM
- GPU · $1.85 / h
- $4.13
- 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.
Massed Compute
No stop; terminating deletes all data.
GPUBank
Only the persistent disk, $0.03 per GB-month. Compute, IPv4 and scratch cost nothing.
04Where the GPUs run
1 country against 4, and where they are.
Countries where Massed Compute 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.
United States
Paris
Frankfurt
Ashburn
Singapore
Massed Compute1
- United States
GPUBank4
Paris
Frankfurt- Ashburn
Singapore
Massed Compute · 1 countryGPUBank · 4 regions
05Paying and signing up
How each one takes your money.
Massed Compute
Payments and sign-up
Payment methods
No cryptoMinimum top-up $10 first top-up
To start
- 1Email and password
- 2Email verification
- 3Card (Stripe)
- 4Age 18+
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.
| Massed Compute | GPUBank | |
|---|---|---|
| SOC 2 | Type II | Not published |
| ISO 27001 | Not published | Not published |
| HIPAA | Listed, report not public | No |
| GDPR DPA | DPA available | Published DPA, applies automatically |
| Availability target | 99.999 % claimed, no SLA document | 99.9 % monthly, per region |
| Status page | None found | Public status page |
| Support | Discord, contact form, email | 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 Massed Compute pricing page captured on 29 September 2026, and GPUBank's live board. Cookie banners and chat bubbles are closed; nothing else is edited.
- 1Prices are shown per configuration, as a total for 1x, 2x, 4x or 8x, not per GPU.
- 2The home page says billing from $0.43+/hr, while the cheapest GPU listed is the A30 at $0.35/hr.
- 38x H100 (80GB) and 8x L40 show a Request button instead of a price.
- 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
28 points, read off both sites.
GPUBank aheadMassed Compute aheadEven
| Criterion | GPUBank | Massed Compute |
|---|---|---|
| Price and catalogue | ||
| H100 SXM, per GPU-hour | $1.85, all-in | $2.89 (1 GPU), $3.14 (8-GPU) |
| Cheapest GPU-hour | $0.23, RTX 4090 | $0.35, A30 |
| GPU models listed | 10 | 19Ahead |
| What the GPU price includes | vCPU, RAM, NVMe, IPv4, traffic | vCPU, RAM, local storage, egress |
| Billing | ||
| Billing granularity | Per second | Per minute, per the docs |
| Minimum chargeThe conflict is unresolved on the site. | None | 1 hour on marketing pages; none in docs |
| Stopped instance | Disk only | No stop; terminating deletes all data |
| Persistent storageThe docs suggest running a CPU VM as an NFS server for persistent data, at extra cost. | $0.03 per GB-month | Local disk bundled; no volume product |
| 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 | Bitcoin only, no auto-recharge |
| Card payments | No | Visa, Mastercard, Amex, Discover; PayPalAhead |
| Minimum deposit | $25 | $10 first top-upAhead |
| Platform | ||
| API and CLI | REST API, CLI, Python SDK | REST API and MCP server; no CLI found |
| Terraform | Official provider | Claimed, not in official repositories |
| Serverless GPUs | No | No |
| Managed Kubernetes | No | None documented |
| Ready-made templates | 25 | PyTorch, vLLM, ComfyUI, Ollama, JupyterLab |
| Interruptible or spot | Yes, 40 % off | No |
| GPUs per instance | Up to 8 | Up to 8, clusters by quoteAhead |
| Multi-node interconnect | Not offered | NVLink variants; InfiniBand on clustersAhead |
| Network and regions | ||
| Regions | 4 | US only, region not selectable |
| Countries | France, Germany, US, Singapore | United States |
| EU regions | Paris, Frankfurt | None |
| Trust and support | ||
| Availability targetThe Terms state the services cannot be guaranteed to be available at all times. | 99.9 % monthly | 99.999 % claimed, no SLA document |
| Compliance reports | None published | SOC 2 Type II and HIPAA claimedAhead |
| Public status page | Yes | None found |
| Support | Tickets from a $25 balance | Discord, contact form, email |
09The right fit
Different needs. A clear choice.
What Massed Compute does better
- 19 GPU models, including A30, RTX A5000, L40 and RTX 6000 Ada
- ThinLinc remote desktop included on every VM
- Cards through Stripe, with PayPal listed in the Terms
- Quoted bare metal and InfiniBand clusters of 8-GPU nodes
- SOC 2 Type II and HIPAA claims, with a published DPA
Choose GPUBank if
- H100 SXM at $1.85 per GPU-hour, against $2.89 to $3.14
- Stop an instance and keep its disk at $0.03 per GB-month
- Four regions in France, Germany, the US and Singapore
- Per-second billing, no one-hour minimum
- Bitcoin, USDT, Ethereum and more, with interruptible and reserved rates
Moving from Massed Compute takes an afternoon.
Massed Compute has no volumes, snapshots or image export, and termination deletes all data. Egress is free, so copy data off before you terminate.
- 1Launch the same GPU on GPUBank with a persistent disk in Ashburn or the region you need.
- 2Copy your data from the Massed Compute VM with rsync or SCP while it is still running.
- 3Recreate the environment from a GPUBank template or your Docker image, then terminate the old VM.
# 1. a GPU and a disk on GPUBank gpubank launch --gpu h100-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.
Yes. Massed Compute lists the H100 SXM5 at $2.89 per GPU-hour on one GPU and $3.14 per GPU on its 8-GPU VM. GPUBank charges $1.85 per GPU-hour for an H100 SXM, with vCPU, RAM, NVMe, IPv4 and traffic included.
No. There is no stop action and no volume product, and terminating a VM deletes all its data. On GPUBank a stopped instance keeps its persistent disk for $0.03 per GB-month, and the disk can move between regions.
The pricing page does not say. It lists H100 (80GB) at $2.73 per GPU-hour separately from the H100 SXM5 and H100 NVL, so this comparison treats it as PCIe. GPUBank rents the H100 PCIe at $1.35 per GPU-hour.
Only Bitcoin, paid directly to a wallet by QR code, and it cannot auto-recharge: you get an email reminder, and instances are terminated if the balance is not topped up. GPUBank takes Bitcoin, USDT, Ethereum and more than a dozen other coins.
In the United States only. The docs say data centers across the United States, no city is published, and the API tells users to set the region to any. GPUBank runs in Paris, Frankfurt, Ashburn and Singapore.
Without. Every price on this page excludes VAT and sales tax, for Massed Compute 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 Massed Compute was read on its own pages or official documentation. Prices last checked on 29 September 2026.
- 01vm.massedcompute.com/pricingAll GPU prices and configurations
- 02massedcompute.com/Product lines, fleet list, API/MCP/SDK claims, compliance badges, EST. 2021, region sample
- 03massedcompute.com/products/on-demand/99.999% uptime claim, Tier III, hourly billing, fleet
- 04massedcompute.com/products/bare-metal/Bare metal setup time, support channel
- 05massedcompute.com/products/gpu-clusters/Cluster node specs, InfiniBand, node counts
- 06massedcompute.com/products/inventory-api/API, SDK and Terraform claims, wholesale pricing mention
- 07massedcompute.com/products/localmetal/On-prem managed offer
- 08massedcompute.com/about-us/Hardware ownership, compliance claims
- 09massedcompute.com/nvidia-partner/Partnership, GPU lineup
- 10massedcompute.com/faq-answers/GPU generations offered incl. H100 PCIe (AI-generated page)
- 11massedcompute.com/contact/Support channels
- 12massedcompute.com/affiliate-program/Referral 3%, affiliate 10%
- 13massedcompute.com/legal/subprocessors/Stripe, AWS and other processors
- 14massedcompute.com/legal/dpa/GDPR DPA, US processing, SOC 2/HIPAA report mention
- 15app.termly.io/policy-viewer/policy.html?policyUUID=a9b12632-9939-4fb4-b7fb-86b695b17c50Legal entity, Nevada address, payment methods, non-refundable purchases, availability disclaimer, 18+
- 16vm-docs.massedcompute.com/docs/introSignup flow, email verification
- 17vm-docs.massedcompute.com/docs/billing/overviewPer-minute debit
- 18vm-docs.massedcompute.com/docs/billing/adding-payment-informationStripe card portal
What we could not verify (11)
- Form factor of 'H100 (80GB)' and 'A100 (80GB)' is not stated on the pricing page (H100 mapped to PCIe on the basis of the AI-generated advisor page; A100 left as 'other')
- VRAM of the B300 SXM6 and B200 SXM6 configurations and of the unlabelled 'H100 NVL' rows is not shown
- Unit of the storage column (GB implied, not stated) and whether it is NVMe
- Whether there is a one-hour minimum charge (marketing) or pure per-minute billing (docs)
- Region names and data center cities: only 'us-central-3' appears in a home-page code sample; docs say data centers across the United States
- Public IPv4 per instance not explicitly stated
- Reserved/commitment, bare metal, cluster and LocalMetal prices are quote-only
- Existence of the Terraform provider and Python/Node.js/Go SDKs claimed on the home page could not be confirmed in official repositories
- Wire transfer/invoicing not published; PayPal appears in the Terms but not in the billing docs
- Funding and headcount not published
- CPU VM prices (mentioned on several pages) were not shown on the captured pricing page
In short
GPUBank vs Massed Compute, in facts.
The essentials of this comparison in one table, for quick readers and AI assistants. Prices last checked on 29 September 2026.
Read as Markdown- H100 SXM, per GPU-hour
- Massed Compute: $2.89 (1 GPU), $3.14 (8-GPU)
GPUBank: $1.85, all-in - Cheapest GPU-hour
- Massed Compute: $0.35, A30
GPUBank: $0.23, RTX 4090 - Billing granularity
- Massed Compute: Per minute, per the docs
GPUBank: Per second - Stopped instance
- Massed Compute: No stop; terminating deletes all data
GPUBank: Disk only - Persistent storage
- Massed Compute: Local disk bundled; no volume product
GPUBank: $0.03 per GB-month - Egress traffic
- Massed Compute: Included
GPUBank: Included - Crypto payments
- Massed Compute: Bitcoin only, no auto-recharge
GPUBank: Bitcoin, USDT, Ethereum and 12+ more - Card payments
- Massed Compute: Visa, Mastercard, Amex, Discover; PayPal
GPUBank: No - Minimum deposit
- Massed Compute: $10 first top-up
GPUBank: $25 - GPU models listed
- Massed Compute: 19
GPUBank: 10 - Regions
- Massed Compute: US only, region not selectable
GPUBank: 4 - Serverless GPUs
- Massed Compute: No
GPUBank: No - Availability target
- Massed Compute: 99.999 % claimed, no SLA document
GPUBank: 99.9 % monthly - Compliance reports
- Massed Compute: SOC 2 Type II and HIPAA claimed
GPUBank: None published - Support
- Massed Compute: Discord, contact form, email
GPUBank: Tickets from a $25 balance - Headquarters
- Massed Compute: Las Vegas, United States
GPUBank: Tallinn, Estonia
How this page was made
Every price and policy of Massed Compute 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. Massed Compute, its logo and its website belong to their owner and are shown only to identify the service compared. GPUBank is not affiliated with Massed Compute.
