GPUBank vs Oblivus
Oblivus is a small GPU cloud run by Skyportal Inc., a Delaware company, that rents KVM virtual machines with 1 to 8 GPUs from Montreal and Oslo, billed per minute from a prepaid balance. Its pricing page lists the H100 SXM5 at $4.31 per GPU-hour, while its docs quote $2.00 as of December 2024; GPUBank charges $1.85. On 2026-09-29 every Hopper and Blackwell SKU at Oblivus showed zero stock.
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
Oblivus
United States · since 2022
H100 SXM
$4.31 / h
- Top up from $5, against $25 at GPUBank
- Cards, bank transfer and crypto through Oblivus Pay
- Reservations of 1 to 36 months, A100 PCIe from $0.95
Oblivus
GPU cloud
28 points compared33 of their pages readprices last checked on 29 September 2026page updated 2 November 2026
The verdict
Low top-ups and more payment options at Oblivus. Lower prices and GPUs in stock at GPUBank.
Oblivus suits small teams that want A100, L40, RTX A6000 or RTX 4090 VMs from a $5 top-up, paid by card, bank transfer or crypto, with Windows available. Its prices conflict across the pricing page, the docs and its own comparison table, and its Hopper and Blackwell stock was zero on the check date. For the GPUs both of us list, GPUBank costs 7 to 64 % less per GPU-hour even against the lowest Oblivus figure. GPUBank bills by the second, includes IPv4 and offers a CLI, Python SDK, Terraform and templates, which Oblivus does not document.
OblivusGPUBank 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 | Oblivus per hour | GPUBank per hour | Difference | 100 h at Oblivus | 100 h at GPUBank |
|---|---|---|---|---|---|
 H200 SXM141 GB |
$3.99 | $2.49 | −38 % | $399.00 | $249.00 |
 H100 SXM80 GB |
$4.31 | $1.85 | −57 % | $431.00 | $185.00 |
 H100 PCIe80 GB |
$3.50 | $1.35 | −61 % | $350.00 | $135.00 |
 A100 SXM80 GB |
$1.57 | $0.95 | −39 % | $157.00 | $95.00 |
 RTX 409024 GB |
$0.64 | $0.23 | −64 % | $64.00 | $23.00 |
The table uses the list prices on Oblivus's pricing page. Its docs quote older rates dated December 2024, such as $2.00 for the H100 SXM. Hopper and Blackwell stock was zero on 2026-09-29.
A fine-tune, 40 h on an H100 SXM
Oblivus
$172.40
GPUBank
$74.00
Saves $98.40 at GPUBank (−57 %)
Image generation, 200 h on an RTX 4090
Oblivus
$128.00
GPUBank
$46.00
Saves $82.00 at GPUBank (−64 %)
A long-context job, 40 h on an H200
Oblivus
$159.60
GPUBank
$99.60
Saves $60.00 at GPUBank (−38 %)
02Their catalogue
Every GPU Oblivus lists.
Oblivus lists 14 GPU models. The 8 below are the ones recorded in our check, each at its lowest public price per GPU-hour. 5 of them are also rented at GPUBank (in blue).
- H200141 GB · on demand$3.99
- H100 SXM80 GB · on demand$4.31
- H100 PCIe80 GB · on demand$3.50
- A100 SXM 80GB80 GB · on demand$1.57
A100 PCIe 80GB80 GB · on demand$1.47
L40on demand$1.05
RTX A6000on demand$0.55- RTX 4090on demand$0.64
At GPUBank
vCPU, RAM, NVMe and traffic included. All prices
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.
Oblivus
2 h 14 m 03 s on H100 SXM
- GPU · $4.31 / h
- $9.63
- vCPU and RAM
- Included
- Storage
- $0.11 per GB-month (NVMe)
- Egress
- Included
- Public IPv4
- $0.0055 per hour, running or stopped
- Billing granularity
- Per minute
- Minimum charge
- Not published
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.
Oblivus
Disk and IPv4 only; IP can be detached.
GPUBank
Only the persistent disk, $0.03 per GB-month. Compute, IPv4 and scratch cost nothing.
04Where the GPUs run
2 countries against 4, and where they are.
Countries where Oblivus 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.
Canada
Norway
Paris
Frankfurt
Ashburn
Singapore
Oblivus2
Canada
Norway
GPUBank4
Paris
Frankfurt
Ashburn
Singapore
Oblivus · 2 countriesGPUBank · 4 regions
05Paying and signing up
How each one takes your money.
Oblivus
Payments and sign-up
Payment methods
Cards
Bank transfer
CryptoCrypto: Yes, via Oblivus Pay; coins not listed
Minimum top-up $5
To start
- 1First name
- 2Last name
- 3Email and password
- 4Country
- 5Captcha
Billing modelPrepaid balance
GPUBank
Payments and sign-up
Payment methods
No cardsA 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.
| Oblivus | GPUBank | |
|---|---|---|
| SOC 2 | Not published | Not published |
| ISO 27001 | Data centers only | Not published |
| HIPAA | Data centers only | No |
| GDPR DPA | Not published | Published DPA, applies automatically |
| Availability target | 99.995 % in docs FAQ, no SLA document | 99.9 % monthly, per region |
| Status page | status.oblivus.com | Public status page |
| Support | Tickets and email; priority for business | 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 Oblivus 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 page lists H100 PCIe at $3.5/hr, while its own comparison table uses $1.98/hr.
- 2The home page says starting at only $0.12/hr, but no GPU at that price is listed anywhere.
- 3A site banner says RTX 5090s are available on demand, while the availability table shows 0.
- 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 aheadOblivus aheadEven
| Criterion | GPUBank | Oblivus |
|---|---|---|
| Price and catalogue | ||
| H100 SXM, per GPU-hourThe docs rate is dated December 2024. All H100 variants showed zero stock on 2026-09-29. | $1.85, all-in | $4.31 pricing page; $2.00 in docs |
| Cheapest GPU-hour | $0.23, RTX 4090 | $0.55, RTX A6000 |
| GPU models listedCounted across all official pages; several have no price or no stock. | 10 | 14Ahead |
| What the GPU price includes | vCPU, RAM, NVMe, IPv4, traffic | vCPU, RAM, NVMe; IPv4 extra |
| Billing | ||
| Billing granularity | Per second | Per minute |
| Minimum charge | None | Not published |
| Stopped instance | Disk only | Disk and IPv4 only; IP can be detached |
| Persistent storage | $0.03 per GB-month | $0.11 per GB-month (NVMe) |
| Egress traffic | Included | Included |
| Public IPv4 | Included | $0.0055 per hour, running or stopped |
| Billing model | Prepaid balance | Prepaid balance |
| Payment and sign-up | ||
| Crypto payments | Bitcoin, USDT, Ethereum and 12+ more | Yes, via Oblivus Pay; coins not listed |
| Card payments | No | Cards and bank transferAhead |
| Minimum deposit | $25 | $5Ahead |
| Platform | ||
| API and CLI | REST API, CLI, Python SDK | REST API; no CLI or SDK documented |
| Terraform | Official provider | None documented |
| Serverless GPUs | No | No |
| Managed Kubernetes | No | None documented |
| Ready-made templates | 25 | OS images only, Ubuntu and Windows |
| Interruptible or spotThe pricing-page FAQ says no spot instances are offered. | Yes, 40 % off | One A100 PCIe spot SKU, above on-demand |
| GPUs per instance | Up to 8 | Up to 8 |
| Multi-node interconnect | Not offered | InfiniBand up to 3.2 Tbps on HGXAhead |
| Network and regions | ||
| RegionsA Texas site shows zero stock; single-tenant Boston and Dallas sites are mentioned in the FAQ. | 4 | 2 public sites, Montreal and Oslo |
| Countries | France, Germany, US, Singapore | Canada, Norway |
| EU regions | Paris, Frankfurt | Oslo, Norway (outside the EU) |
| Trust and support | ||
| Availability target | 99.9 % monthly | 99.995 % in docs FAQ, no SLA document |
| Compliance reports | None published | Facility-level ISO 27001, SOC 2 only |
| Public status page | Yes | status.oblivus.com |
| Support | Tickets from a $25 balance | Tickets and email; priority for business |
09The right fit
Different needs. A clear choice.
What Oblivus does better
- Top up from $5, against $25 at GPUBank
- Cards, bank transfer and crypto through Oblivus Pay
- Reservations of 1 to 36 months, A100 PCIe from $0.95
- Windows VMs with RDP, and L40 and RTX A6000 on demand
- InfiniBand up to 3.2 Tbps on HGX nodes
Choose GPUBank if
- H100 SXM at $1.85 per GPU-hour, against $4.31 on their pricing page
- H100 and H200 capacity on sale, where Oblivus showed zero stock
- Per-second billing, and IPv4 included instead of $0.0055 per hour
- CLI, Python SDK, Terraform provider, templates and Docker images
- Stopped disk at $0.03 per GB-month, against $0.11
Moving from Oblivus takes an afternoon.
Oblivus charges nothing for egress, so copying data out is free. Network disks attach only within one data center and no image export is documented, so move files, not disks.
- 1Launch the same GPU on GPUBank with a persistent disk in the region you need.
- 2Copy your data from the Oblivus VM with rsync or scp over its public network.
- 3Recreate your environment from a GPUBank template or your Docker image, then delete the Oblivus VM and its IP.
# 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. The Oblivus pricing page lists the H100 SXM5 at $4.31 per GPU-hour, and its docs quote $2.00 as of December 2024. GPUBank charges $1.85 per GPU-hour for an H100 SXM, with IPv4 and traffic included.
Oblivus publishes three price sets that disagree: the pricing page, a docs FAQ dated December 2024 and a comparison table on the pricing page. This page treats the pricing page as primary and shows the docs figure where it is the only or lowest one.
Not on the check date. On 2026-09-29 the Oblivus availability page showed zero stock for B200, H200, every H100 variant and A100 NVLINK. What was deployable was A100 80GB PCIe, L40, RTX A6000 and RTX 4090, all in Montreal.
Yes. Oblivus takes crypto through its in-house Oblivus Pay, without listing the coins, plus cards and bank transfer, from a $5 top-up. GPUBank takes Bitcoin, USDT, Ethereum and more than a dozen other coins from $25, and no cards.
At Oblivus a stopped VM pays NVMe at $0.11 per GB-month plus its IPv4 at $0.0055 per hour, unless you detach the IP. On GPUBank a stopped instance pays only its persistent disk, at $0.03 per GB-month.
Without. Every price on this page excludes VAT and sales tax, for Oblivus 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 Oblivus was read on its own pages or official documentation. Prices last checked on 29 September 2026.
- 01oblivus.com/Banner, stock tiles, billing/payment/account FAQ
- 02oblivus.com/pricing/On-demand and reservation GPU prices, storage, IP, egress, comparison table, payment and billing FAQ
- 03oblivus.com/availability/Live stock by region, B200, RTX 5090, RTX PRO6000 SE, spot SKUs, TXS1
- 04oblivus.com/datacenters/MON1, OSL1, 7 single-tenant + 2 multi-tenant, facility compliance
- 05oblivus.com/cloud/Per-minute billing, KVM, bare metal on reservation, OS options, ISO claims, 5 USD top-up by card/bank/crypto
- 06oblivus.com/storage/0.11 USD/GB-month NVMe, 3x replication
- 07oblivus.com/networking/Unlimited bandwidth, 3.2 Tbps InfiniBand, per-GPU public port
- 08oblivus.com/restrictions/Max 8 GPUs per instance, per-GPU vCPU/RAM/storage
- 09oblivus.com/ourdifference/Crypto via Binance (2022) then Oblivus Pay (2024)
- 10oblivus.com/aboutus/Company history 2022-2024
- 11oblivus.com/reserve/H200 reservations, GB200 waitlist
- 12oblivus.com/referral/3% commission, 2% cashback
- 13oblivus.com/terms/Skyportal Inc., Stripe/Oblivus Pay processors, non-refundable funds, entity registration needs authorization
- 14oblivus.com/privacy/Legal entity Skyportal Inc., Delaware
- 15docs.oblivus.com/Regions, facility compliance table
- 16docs.oblivus.com/instances/instance-types/Docs GPU prices incl. RTX 5090, RTX PRO6000 SE, SKY H100/H200
- 17docs.oblivus.com/support/faq/Prices as of Dec 2024, regions incl. TXS1/Boston/Dallas, 99.995% SLA, support tiers, Auto Top-Up threshold, business account criteria
- 18docs.oblivus.com/support/changelog/RTX 5090 (OSL1), RTX PRO6000 SE (MON1), H200, Hyperstack API integration
What we could not verify (11)
- Which price is current for H100 SXM5/NVLINK/PCIe and H200: pricing page (4.312/3.64/3.5/3.99) vs docs (2.00/2.08/1.98/3.42); pricing page treated as primary
- RTX 5090 and RTX PRO6000 SE prices only in docs (0.9 and 2.0 USD/h); VRAM of RTX PRO6000 SE not stated
- VRAM of L40, RTX 4090, RTX A6000 not stated on Oblivus pages (left null)
- Reservation price per term (1-36 months) not published; only 'starting from' floors
- B200 SXM listed in the availability table without a price; RTX A5000 and RTX A4000 named in the docs FAQ without a price
- Coins/networks accepted by Oblivus Pay not published
- Content of the individual 'straightforward verification process' not documented; business accounts may be asked for documents
- City of TXS1 and locations of the 7 single-tenant sites not published
- No SLA terms document; 99.995% appears only in the docs FAQ
- CLI, SDK, Terraform, Kubernetes, Jupyter not documented (null)
- Funding not published
In short
GPUBank vs Oblivus, 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
- Oblivus: $4.31 pricing page; $2.00 in docs
GPUBank: $1.85, all-in - Cheapest GPU-hour
- Oblivus: $0.55, RTX A6000
GPUBank: $0.23, RTX 4090 - Billing granularity
- Oblivus: Per minute
GPUBank: Per second - Stopped instance
- Oblivus: Disk and IPv4 only; IP can be detached
GPUBank: Disk only - Persistent storage
- Oblivus: $0.11 per GB-month (NVMe)
GPUBank: $0.03 per GB-month - Egress traffic
- Oblivus: Included
GPUBank: Included - Crypto payments
- Oblivus: Yes, via Oblivus Pay; coins not listed
GPUBank: Bitcoin, USDT, Ethereum and 12+ more - Card payments
- Oblivus: Cards and bank transfer
GPUBank: No - Minimum deposit
- Oblivus: $5
GPUBank: $25 - GPU models listed
- Oblivus: 14
GPUBank: 10 - Regions
- Oblivus: 2 public sites, Montreal and Oslo
GPUBank: 4 - Serverless GPUs
- Oblivus: No
GPUBank: No - Availability target
- Oblivus: 99.995 % in docs FAQ, no SLA document
GPUBank: 99.9 % monthly - Compliance reports
- Oblivus: Facility-level ISO 27001, SOC 2 only
GPUBank: None published - Support
- Oblivus: Tickets and email; priority for business
GPUBank: Tickets from a $25 balance - Headquarters
- Oblivus: United States
GPUBank: Tallinn, Estonia
How this page was made
Every price and policy of Oblivus 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. Oblivus, its logo and its website belong to their owner and are shown only to identify the service compared. GPUBank is not affiliated with Oblivus.
