GPUBank vs AWS (Amazon EC2)
Amazon EC2 is the AWS compute service, with 18 GPU and accelerator models across 39 regions, billed per second. Prices here are us-east-1 on-demand rates per instance divided by the GPU count: the H100 costs $6.88 per GPU-hour, against $1.85 at GPUBank. AWS offers scale, managed services, a formal SLA and a full compliance portfolio; GPUBank costs less and includes IPv4 and all traffic in the GPU price.
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
AWS (Amazon EC2)
Seattle, United States
H100 SXM
$6.88 / h
- 18 accelerator models, from T4 and L4 to B300 and GB200
- 39 regions and 124 availability zones
- 99.99% region-level and 99.5% instance-level SLA with credits
AWS (Amazon EC2)
Hyperscaler
United States26 points compared18 of their pages readprices last checked on 29 September 2026page updated 2 November 2026
The verdict
Scale, services and compliance at AWS. A fraction of the price, all-in, at GPUBank.
AWS fits teams that already run on AWS and need GPUs next to their data, inside their compliance scope, with EKS, SageMaker and UltraClusters around them. On the five GPUs both of us list, GPUBank costs 66 to 73 % less per GPU-hour, and AWS adds EBS storage, $0.005 per IPv4 address-hour and $0.09 per GB of egress after 100 GB a month. Capacity Blocks bring the AWS H100 to $5.19 per GPU-hour, still above GPUBank. AWS is paid by card or direct debit; GPUBank is topped up in crypto from $25.
AWS (Amazon EC2)GPUBank 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 | AWS (Amazon EC2) per hour | GPUBank per hour | Difference | 100 h at AWS (Amazon EC2) | 100 h at GPUBank |
|---|---|---|---|---|---|
 B200 SXM180 GB |
$14.24 | $4.15 | −71 % | $1,424.00 | $415.00 |
 H200 SXM141 GB |
$7.91 | $2.49 | −69 % | $791.00 | $249.00 |
 H100 SXM80 GB |
$6.88 | $1.85 | −73 % | $688.00 | $185.00 |
 RTX PRO 600096 GB |
$3.36 | $1.15 | −66 % | $336.00 | $115.00 |
 L40S48 GB |
$1.86 | $0.55 | −70 % | $186.00 | $55.00 |
A fine-tune, 40 h on an H100 SXM
AWS (Amazon EC2)
$275.20
GPUBank
$74.00
Saves $201.20 at GPUBank (−73 %)
An inference box, a month, 720 h, on an L40S
AWS (Amazon EC2)
$1,339.20
GPUBank
$396.00
Saves $943.20 at GPUBank (−70 %)
A long-context job, 40 h on an H200
AWS (Amazon EC2)
$316.40
GPUBank
$99.60
Saves $216.80 at GPUBank (−69 %)
02Their catalogue
Every GPU AWS (Amazon EC2) lists.
18 GPU models listed by AWS (Amazon EC2), each at its lowest public price per GPU-hour. 5 of them are also rented at GPUBank (in blue).
- Blackwell and Hopper5
GB200185 GB · reserved$10.58- B300268 GB · on demand$17.80
- B200179 GB · on demand$14.24
- H200141 GB · on demand$7.91
- H100 SXM80 GB · on demand$6.88
- Ampere, Ada and older data-centre9
- A100 40GB40 GB · on demand$2.74
- A100 80GB80 GB · reserved$2.21
- Ampere, Ada and older data-centre continued9
- L40S44 GB · on demand$1.86
L422 GB · on demand$0.80
A10G22 GB · on demand$1.01
T416 GB · on demand$0.53
V10016 GB · on demand$3.06- M608 GB · on demand$1.14
- T4g16 GB · on demand$0.42
- Workstation2
- RTX PRO 600096 GB · on demand$3.36
- RTX PRO 4500 Blackwell32 GB · on demand$2.52
- AMD and other accelerators2
Gaudi HL-20532 GB · on demand$1.64- Radeon Pro V5208 GB · on demand$0.38
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.
AWS (Amazon EC2)
2 h 14 m 03 s on H100 SXM
- GPU · $6.88 / h
- $15.37
- vCPU and RAM
- Included
- Storage
- $0.08 per GB-month, EBS gp3
- Egress
- $0.09 per GB after 100 GB a month
- Public IPv4
- $0.005 per address-hour
- Billing granularity
- Per second
- Minimum charge
- 60 seconds per launch
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.
AWS (Amazon EC2)
EBS and public IPv4 keep billing.
GPUBank
Only the persistent disk, $0.03 per GB-month. Compute, IPv4 and scratch cost nothing.
04Where the GPUs run
6+ countries against 4, and where they are.
Countries where AWS (Amazon EC2) 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
Japan
India
United Kingdom
Sweden
Brazil
Paris
Frankfurt
Ashburn
Singapore
AWS (Amazon EC2)6
- United States
Japan
India
United Kingdom
Sweden
Brazil
GPUBank4
Paris
Frankfurt- Ashburn
Singapore
AWS (Amazon EC2) · 6 countriesGPUBank · 4 regions
05Paying and signing up
How each one takes your money.
AWS (Amazon EC2)
Payments and sign-up
Payment methods
Cards
ACH direct debit
No cryptoMinimum top-up Not published
To start
- 1Email and password
- 2Card
- 3Billing address
- 4Phone
- 5Additional verification for India-based accounts
Billing modelPostpaid; card or direct debit
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.
| AWS (Amazon EC2) | GPUBank | |
|---|---|---|
| SOC 2 | Yes | Not published |
| ISO 27001 | Yes | Not published |
| HIPAA | Yes | No |
| GDPR DPA | DPA available | Published DPA, applies automatically |
| Availability target | 99.99% region, 99.5% instance, credits | 99.9 % monthly, per region |
| Status page | Not published | Public status page |
| Support | Not published | 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 AWS (Amazon EC2) pricing page captured on 29 September 2026, and GPUBank's live board. Cookie banners and chat bubbles are closed; nothing else is edited.
- 1The On-Demand pricing page shows prices only through a JavaScript region and OS selector; the figures come from the JSON feed it loads.
- 2The Capacity Blocks page lists effective hourly rates per instance and per accelerator by region.
- 3The Spot page shows the lowest current Spot price per instance type, updated every 5 minutes.
- 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 aheadAWS (Amazon EC2) aheadEven
| Criterion | GPUBank | AWS (Amazon EC2) |
|---|---|---|
| Price and catalogue | ||
| H100 SXM, per GPU-hourus-east-1 on demand. Capacity Blocks book it from $5.19 per GPU-hour. | $1.85, all-in | $6.88, p5 with 1 or 8 GPUs |
| Cheapest GPU-hour | $0.23, RTX 4090 | $0.38, AMD Radeon Pro V520 |
| GPU models listed | 10 | 18Ahead |
| What the GPU price includesPublic IPv4 and egress are also billed separately. | vCPU, RAM, NVMe, IPv4, traffic | vCPU, RAM, instance NVMe; EBS extra |
| Billing | ||
| Billing granularity | Per second | Per second |
| Minimum charge | None | 60 seconds per launch |
| Stopped instance | Disk only | EBS and public IPv4 keep billing |
| Persistent storage | $0.03 per GB-month | $0.08 per GB-month, EBS gp3 |
| Egress trafficFalls to $0.05 per GB above 150 TB. Inbound is free. | Included | $0.09 per GB after 100 GB a month |
| Public IPv4 | Included | $0.005 per address-hour |
| Billing model | Prepaid balance | Postpaid; card or direct debit |
| Payment and sign-up | ||
| Crypto payments | Bitcoin, USDT, Ethereum and 12+ more | No |
| Card payments | No | Credit and debit cardsAhead |
| Minimum deposit | $25 | Not published |
| Platform | ||
| API and CLI | REST API, CLI, Python SDK | API, CLI, SDKs in many languages |
| Terraform | Official provider | Official provider |
| Serverless GPUs | No | YesAhead |
| Managed Kubernetes | No | EKS, managedAhead |
| Ready-made templates | 25 | Deep Learning AMIs and containers |
| Interruptible or spotOn the check date, p5.48xlarge Spot was above its on-demand price. | Yes, 40 % off | Spot, variable price |
| GPUs per instance | Up to 8 | Up to 8; UltraClusters to 20,000 GPUsAhead |
| Multi-node interconnect | Not offered | 900 GB/s NVSwitch, EFA up to 3,200 Gb/sAhead |
| Network and regions | ||
| RegionsGPU availability differs by region. | 4 | 39Ahead |
| Countries | France, Germany, US, Singapore | US, Japan, India, UK, Sweden, Brazil and moreAhead |
| EU regions | Paris, Frankfurt | Yes, e.g. StockholmAhead |
| Trust and support | ||
| Availability target | 99.9 % monthly | 99.99% region, 99.5% instance, creditsAhead |
| Compliance reports | None published | SOC 1/2/3, ISO 27001, HIPAA BAAAhead |
| Public status page | Yes | Not published |
| SupportSupport plan tiers and prices were not captured. | Tickets from a $25 balance | Not published |
09The right fit
Different needs. A clear choice.
What AWS (Amazon EC2) does better
- 18 accelerator models, from T4 and L4 to B300 and GB200
- 39 regions and 124 availability zones
- 99.99% region-level and 99.5% instance-level SLA with credits
- SOC 1, 2 and 3, ISO 27001 and a HIPAA BAA
- EKS, SageMaker, EFA and UltraClusters up to 20,000 GPUs
Choose GPUBank if
- H100 SXM at $1.85 per GPU-hour, against $6.88 on demand
- IPv4 and all traffic included; AWS bills both
- Single H200, B200 and A100 GPUs; AWS sells these as 8-GPU instances
- Interruptible at a fixed -40 %; AWS Spot varies
- Pay in Bitcoin, USDT and 12+ other coins, no card or phone check
Moving from AWS (Amazon EC2) takes an afternoon.
EBS snapshots and AMIs stay inside AWS, and egress costs $0.09 per GB after the free 100 GB. The AWS pricing page says it offers free egress to eligible customers leaving AWS entirely.
- 1Launch the same GPU on GPUBank with a persistent disk in the region you need.
- 2Copy data from EBS or S3 with rsync or rclone, after checking whether AWS waives egress for your exit.
- 3Run your Deep Learning Container or Docker image on GPUBank, then terminate the instances and release EBS volumes and IPv4 addresses.
# 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. In us-east-1, AWS lists the p5 H100 at $6.88 per GPU-hour on demand, and from $5.19 as a Capacity Block. GPUBank charges $1.85 per GPU-hour, with vCPU, RAM, NVMe, IPv4 and traffic included.
We take the us-east-1 Linux on-demand price of each instance and divide it by its GPU count. AWS prices vary by region. EBS storage, public IPv4 and egress are billed on top of those figures.
Not on demand in us-east-1. The H200 and B200 come as 8-GPU instances; the H100 is the large GPU with a single-GPU size, p5.4xlarge. GPUBank rents every model from 1 to 8 GPUs.
AWS. It publishes a 99.99% region-level and 99.5% instance-level SLA with service credits, and holds SOC 1, 2 and 3, ISO 27001 and a HIPAA BAA. GPUBank targets 99.9 % per region and publishes no SOC 2 or ISO 27001 report.
No. AWS takes credit and debit cards, and direct debit or invoicing for qualifying customers; PayPal is not accepted. Sign-up needs a card, a billing address and a phone check. GPUBank is paid from a USD balance topped up in Bitcoin, USDT, Ethereum and more than a dozen other coins.
Without. Every price on this page excludes VAT and sales tax, for AWS (Amazon EC2) 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 AWS (Amazon EC2) was read on its own pages or official documentation. Prices last checked on 29 September 2026.
- 01aws.amazon.com/ec2/pricing/on-demand/Billing model, data transfer prices, 100 GB free egress
- 02b0.p.awsstatic.com/pricing/2.0/meteredUnitMaps/ec2/USD/current/ec2-ondemand-without-sec-sel/US%20East%20(N.%20Virginia)/Linux/index.jsonHourly on-demand price per instance type
- 03website.spot.ec2.aws.a2z.com/spot.jsus-east-1 Linux Spot prices
- 04aws.amazon.com/ec2/spot/pricing/Spot mechanics, up to 90% off
- 05aws.amazon.com/ec2/capacityblocks/pricing/Capacity Block rates
- 06docs.aws.amazon.com/ec2/latest/instancetypes/ac.htmlGPU model, count, memory per instance type
- 07docs.aws.amazon.com/ec2/latest/instancetypes/pg.htmlP3, G3 GPU specs
- 08aws.amazon.com/ec2/instance-types/p5/NVSwitch, EFA, UltraClusters, instance store
- 09aws.amazon.com/ec2/pricing/Per-second billing, 60 s minimum
- 10aws.amazon.com/ebs/pricing/gp3/gp2 prices
- 11aws.amazon.com/vpc/pricing/Public IPv4 0.005 USD/hour
- 12repost.aws/knowledge-center/accepted-payment-methodsCards, ACH, SEPA, Bacs, PayPal not accepted
- 13repost.aws/knowledge-center/create-and-activate-aws-accountPayment method, phone verification
- 14aws.amazon.com/compute/sla/99.99% / 99.5% SLA
- 15aws.amazon.com/about-aws/global-infrastructure/regions_az/39 regions, 124 AZs
- 16aws.amazon.com/compliance/programs/SOC, ISO, HIPAA BAA, DPA
- 17aws.amazon.com/free/Up to 200 USD credits
- 18aws.amazon.com/agreement/Headquarters
What we could not verify (7)
- p2 (K80) and g3s.xlarge prices exist in the feed but their GPU specs are not on the current docs pages read, so they are excluded
- A100 80GB (p4de) form factor not stated
- Savings Plans and Reserved Instance GPU rates not captured
- Founding year and HQ city not stated on pages read
- Spot prices are volatile; values are a snapshot of the feed on 2026-09-29
- Support plan tiers and prices not captured
- Status page URL not captured
In short
GPUBank vs AWS (Amazon EC2), 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
- AWS (Amazon EC2): $6.88, p5 with 1 or 8 GPUs
GPUBank: $1.85, all-in - Cheapest GPU-hour
- AWS (Amazon EC2): $0.38, AMD Radeon Pro V520
GPUBank: $0.23, RTX 4090 - Billing granularity
- AWS (Amazon EC2): Per second
GPUBank: Per second - Stopped instance
- AWS (Amazon EC2): EBS and public IPv4 keep billing
GPUBank: Disk only - Persistent storage
- AWS (Amazon EC2): $0.08 per GB-month, EBS gp3
GPUBank: $0.03 per GB-month - Egress traffic
- AWS (Amazon EC2): $0.09 per GB after 100 GB a month
GPUBank: Included - Crypto payments
- AWS (Amazon EC2): No
GPUBank: Bitcoin, USDT, Ethereum and 12+ more - Card payments
- AWS (Amazon EC2): Credit and debit cards
GPUBank: No - Minimum deposit
- AWS (Amazon EC2): Not published
GPUBank: $25 - GPU models listed
- AWS (Amazon EC2): 18
GPUBank: 10 - Regions
- AWS (Amazon EC2): 39
GPUBank: 4 - Serverless GPUs
- AWS (Amazon EC2): Yes
GPUBank: No - Availability target
- AWS (Amazon EC2): 99.99% region, 99.5% instance, credits
GPUBank: 99.9 % monthly - Compliance reports
- AWS (Amazon EC2): SOC 1/2/3, ISO 27001, HIPAA BAA
GPUBank: None published - Support
- AWS (Amazon EC2): Not published
GPUBank: Tickets from a $25 balance - Headquarters
- AWS (Amazon EC2): Seattle, United States
GPUBank: Tallinn, Estonia
How this page was made
Every price and policy of AWS (Amazon EC2) 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. AWS (Amazon EC2), its logo and its website belong to their owner and are shown only to identify the service compared. GPUBank is not affiliated with AWS (Amazon EC2).
