gpurank
Compare pricesCompare
esc
  • Nothing named “”. Press Enter to search the offers instead.
↑↓ to move · ↵ to open · esc to close0 entries
Operator · Wilmington, DE · founded 2022

Oblivus

Small US-registered GPU cloud (Skyportal Inc. dba Oblivus) selling KVM GPU VMs billed per minute from Montreal and Oslo, on a prepaid balance funded by card, bank transfer or crypto (Oblivus Pay).

GPU types9with a current offer
Current offers6767 readings on file · last 45 days
H100 from$3.50per GPU-hour, on-demand
Entry price$1.05cheapest GPU-hour listed
Countries27 named regions
Trust Score v0.1
82/100
  • Company & track record73
  • Infrastructure control100
  • Pricing & billing transparency100
  • Reliability signals100
  • Security & compliance27
  • Customer signal—

Scored Sep 29, 2026 from the provider's own pages. Dimensions not yet measured are excluded from the denominator. Criteria ↓

67 current offers →oblivus.com ↗

Outbound link carries a commission. It does not affect this page.

Trust Score · 5 of 6 dimensions measured

The score, criterion by criterion

Each point comes from something the provider publishes, with the page it was read on and the date. Anything not disclosed earns zero.

Company & track record
weight 15%
73 / 100
Founding year published2022oblivus.com · Sep 29, 20264 / 4
Headquarters publishedSkyportal Inc. doing business as Oblivusoblivus.com · Sep 29, 20263 / 3
Legal entity namedSkyportal Inc. doing business as Oblivusoblivus.com · Sep 29, 20264 / 4
Funding or ownership disclosedPrivate; funding not published. About page: Oblivus Limited established March 2022, operations restructured under Oblivus, Inc. (USA) in May 2024; site footer and console still say 'Oblivus Limited'.oblivus.com · Sep 29, 20260 / 4
Infrastructure control
weight 15%
100 / 100
Capacity model publishedoperates or rents dedicated capacityoblivus.com · Sep 29, 202610 / 10
Regions published7 regions or countriesoblivus.com · Sep 29, 20265 / 5
Pricing & billing transparency
weight 20%
100 / 100
Public prices without an accountyes (read on the pricing page)oblivus.com · Sep 29, 20265 / 5
Billing granularity publishedper minuteoblivus.com · Sep 29, 20264 / 4
Egress policy publishedincluded (ingress and egress free, 'unlimited bandwidth'; 10 to 100 Gbps public port depending on GPU model)oblivus.com · Sep 29, 20264 / 4
Stopped-instance billing publishedStorage and the assigned IPv4 only (NVMe 0.000154 USD/GB-hour + IP 0.0055 USD/h; console shows 0.0209 USD/h stopped for a 100 GB root disk). The IP can be detached while stopped to avoid the IP charge.oblivus.com · Sep 29, 20264 / 4
Public IPv4 policy publishedextra: 0.0055 USD/h (4 USD/month) per IP, billed while running and while stopped; detachableoblivus.com · Sep 29, 20263 / 3
Reliability signals
weight 20%
100 / 100
Status pagehttps://status.oblivus.comoblivus.com · Sep 29, 20266 / 6
SLA published99.995% uptime SLA stated for Tier-4 data centers (docs FAQ); no SLA document published; home page claims '%99.99 Uptime'oblivus.com · Sep 29, 20266 / 6
SLA with a numeric uptime commitment99.995% uptime SLA stated for Tier-4 data centers (docs FAQ); no SLA document published; home page claims '%99.99 Uptime'oblivus.com · Sep 29, 20263 / 3
Uptime history publishedStatus page: MON1, OSL1, Console, API at 100% over 90 days (2026-09-29)oblivus.com · Sep 29, 20265 / 5
Observed availability (30 days)not measured yet—
Security & compliance
weight 15%
27 / 100
SOC 2not published for Oblivus; MON1 and OSL1 facilities stated SOC 2 Type 2oblivus.com · Sep 29, 20262 / 8
ISO 27001Facility level only: MON1 and OSL1 stated ISO 27001; compute page cites ISO/IEC 27001, 27017, 27018, 27701 data centersoblivus.com · Sep 29, 20261 / 4
HIPAA or GDPR DPAFacility level: MON1 and OSL1 stated HIPAA compliant; no BAA offer publishedoblivus.com · Sep 29, 20261 / 3
Customer signal
weight 15%
not measured
Aggregated public reviewsnot collected yet—
provisional v0.1: evidence from the provider's own pages; dimensions not yet measured are excluded from the denominatorFull method →
Prices by GPU · 4 GPU types with a current offer

What Oblivus charges, against the market

Cheapest current on-demand true cost here, next to the market median for the same GPU. Negative means cheaper than the market.

GPUCheapest hereMarket medianvs marketCurrent offersModes
H100$3.50$3.17+11%6on-demandreservedH100 prices
A100 80GB$1.47$1.99−26%4on-demandreservedA100 80GB prices
RTX 4090$0.64$0.44+45%2on-demandreservedRTX 4090 prices
L40$1.05$0.89+18%2on-demandreservedL40 prices
All offers · 67 readings on file

Every Oblivus offer, with its receipt

Current offers (last 45 days) ranked by true cost, all rental modes, each with its read date. Older readings stay listed below, in grey.

GPUConfigurationModeRegionStockListTrue costRead
RTX 40908× RTX 4090ReservedOblivus data centresUnavailable$0.44$0.4421:10 UTCfreshReceipt
RTX 40908× RTX 4090On-demandOblivus data centresUnavailable$0.64$0.6421:10 UTCfreshReceipt
L408× L40ReservedOblivus data centresIn stock$0.74$0.7421:10 UTCfreshReceipt
A100 80GB8× A100 80GBReservedOblivus data centresUnavailable$0.95$0.9521:10 UTCfreshReceipt
A100 80GB8× A100 80GBReservedOblivus data centresUnavailable$1.05$1.0521:10 UTCfreshReceipt
L408× L40On-demandOblivus data centresIn stock$1.05$1.0521:10 UTCfreshReceipt
A100 80GB8× A100 80GBOn-demandOblivus data centresUnavailable$1.47$1.4721:10 UTCfreshReceipt
A100 80GB8× A100 80GBOn-demandOblivus data centresUnavailable$1.57$1.5721:10 UTCfreshReceipt
H100PCIE8× H100 PCIEReservedOblivus data centresIn stock$2.56$2.5621:10 UTCfreshReceipt
H100PCIE8× H100 PCIEReservedOblivus data centresUnavailable$2.71$2.7121:10 UTCfreshReceipt
H100SXM8× H100 SXMReservedOblivus data centresUnavailable$2.92$2.9221:10 UTCfreshReceipt
H100PCIE8× H100 PCIEOn-demandOblivus data centresIn stock$3.50$3.5021:10 UTCfreshReceipt
H100PCIE8× H100 PCIEOn-demandOblivus data centresUnavailable$3.64$3.6421:10 UTCfreshReceipt
H100SXM8× H100 SXMOn-demandOblivus data centresUnavailable$4.31$4.3121:10 UTCfreshReceipt
14 fresh offers · prices per GPU-hour · true cost includes the add-ons the provider publishesOpen in the comparison tool →
True cost & billing · read Sep 29, 2026

What the hourly price does not say

Billing granularity, minimums, what a stopped machine costs, traffic, addresses, deposits and identity checks, as published by the provider.

Billing

Billing granularity
per minute
Minimum charge
not published
Stopped instance
Storage and the assigned IPv4 only (NVMe 0.000154 USD/GB-hour + IP 0.0055 USD/h; console shows 0.0209 USD/h stopped for a 100 GB root disk). The IP can be detached while stopped to avoid the IP charge.
Egress
included (ingress and egress free, 'unlimited bandwidth'; 10 to 100 Gbps public port depending on GPU model)
Public IPv4
extra: 0.0055 USD/h (4 USD/month) per IP, billed while running and while stopped; detachable
Billing model
prepaid credits (optional card Auto Top-Up; business accounts can pay invoices)
Minimum deposit
5 USD minimum top-up; Auto Top-Up low-balance threshold minimum 10 USD
Free credits
No free trial; occasional promo codes; referred sign-ups get 2% cashback (up to 25,000 USD); no formal academic credit program

Account

Payment
card · invoice · crypto (yes: via in-house 'Oblivus Pay' (Binance earlier); coins and networks not published)
Sign-up
first name, last name, email, password, country, captcha, email verification, business accounts: application reviewed (up to 2 business days), documents may be requested
Identity check
ID on request
Support
tickets (console), email
Enterprise support
not published
SLA
99.995% uptime SLA stated for Tier-4 data centers (docs FAQ); no SLA document published; home page claims '%99.99 Uptime'
Published uptime
Status page: MON1, OSL1, Console, API at 100% over 90 days (2026-09-29)

Infrastructure

Regions
7 named regions in 2 countriesMON1 (Canada) · OSL1 (Norway) · TXS1 (United States) · Boston (single-tenant · Tier-4) (United States) · Dallas (single-tenant · Tier-4) (United States)
Countries
Canada, Norway
Platform
API, SSH, persistent volumes, snapshots, spot or interruptibleInterconnect: InfiniBand up to 3.2 Tbps on HGX modules; 10-100 Gbps public network per GPU modelClusters: Bare metal only on long-term reservations (contract-size dependent); FAQ mentions Tier-4 single-tenant sites with 3200 Gbps InfiniBand
Storage
volume · 0.000154 USD/GB-hour (0.11 USD/GB-month on the storage page) · NVMe block storage, 3x replicated (Ceph); included in the GPU price while the instance runs, billed alone when stopped. 0.000154 x 720 h = 0.111 USD/GB-month.network · 0.000154 USD/GB-hour · NVMe shared (network) disks, attachable to instances in the same data center, persist after instance deletion.ephemeral · included · Console lists ephemeral NVMe per flavor (e.g. 750 GB for 1x A100 PCIe) as included.
Migration out
VM-based: data sits on local NVMe (root/ephemeral) or network disks; export by copying data out over the free-egress public network (scp/rsync). Network disks persist after instance deletion but attach only within one data center. Images can be replicated between Oblivus data centers; no image export is documented.

Security & compliance

SOC 2
not published for Oblivus; MON1 and OSL1 facilities stated SOC 2 Type 2
ISO 27001
Facility level only: MON1 and OSL1 stated ISO 27001; compute page cites ISO/IEC 27001, 27017, 27018, 27701 data centers
HIPAA
Facility level: MON1 and OSL1 stated HIPAA compliant; no BAA offer published
GDPR / DPA
not published
Other
SOC 1 Type 2 at facility level; HITRUST not held
Legal entity
Skyportal Inc. doing business as Oblivus
Funding
Private; funding not published. About page: Oblivus Limited established March 2022, operations restructured under Oblivus, Inc. (USA) in May 2024; site footer and console still say 'Oblivus Limited'.
Read on the provider's own pages · 33 pages read · facts marked "not published" were looked for and not foundPages read ↓
Assessment · documents read Sep 29, 2026

Oblivus, in short

What stood out when we read the documents: strengths, weaknesses, who it suits, and what we could not find.

Strengths

  • Per-minute billing on a prepaid balance with a 5 USD minimum top-up
  • Crypto accepted through in-house Oblivus Pay, plus card (Revolut/Stripe) and bank transfer
  • Ingress and egress free; public port 10 to 100 Gbps depending on GPU model
  • On-demand RTX A6000 0.55, RTX 4090 0.64, A100 80GB PCIe 1.47 USD/GPU-h
  • Reservations of 1 to 36 months with advertised floors of 0.95 USD/h (A100 PCIe) and 2.919 USD/h (H100 SXM5)
  • Stopped instances billed only for storage and IP; IP can be detached
  • No quotas advertised; up to 8 GPUs per VM; KVM with GPU passthrough; Windows supported

Weaknesses

  • Published prices are internally inconsistent: pricing page H100 SXM5 4.312 / H100 PCIe 3.5 USD/h vs docs 2.00 / 1.98 USD/h and a comparison table at 1.98
  • Hopper and Blackwell stock at 0 in every region on 2026-09-29; 45 GPUs available across multi-tenant sites
  • Public IPv4 billed on top of the 'all-in' price (0.0055 USD/h)
  • Top-ups non-refundable per Terms of Service
  • No CLI, SDK, Terraform, managed Kubernetes or serverless documented; VMs only, no container-image deployment
  • Most docs pages 'last updated 1 year ago'; entity naming split between Oblivus Limited, Oblivus, Inc. and Skyportal Inc.
  • Spot SKU priced above on-demand, and an FAQ states no spot is offered
  • Priority support restricted to business accounts

Ideal for

  • Individuals or small teams wanting A100/L40/A6000/4090 VMs paid by card or crypto from 5 USD
  • Buyers negotiating multi-month reservations who confirm the rate with sales

Not published

  • 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)
  • 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
  • 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
  • Funding not published

Observations

Pricing page shows per-GPU on-demand and 'Reservations starting from' prices without an account; states prices are all-in and multi-GPU cost = hourly price x GPUs

Site-wide banner: 'Latest generation RTX 5090s now available on demand', yet the pricing page lists no RTX 5090 and the availability table shows 0 RTX 5090

Home hero: 'starting at only $0.12/hr'; no GPU at that price is listed anywhere; home GPU tiles quote reservation floors ($0.38, $0.95, $2.56) as 'Starting from'

Pricing-page comparison table uses 1.98/hr for H100 PCIe and 12.56/hr for 8x A100 NVLINK, while the same page lists H100 PCIe at 3.5/hr

Availability page on 2026-09-29: B200 SXM, H200, all H100 variants, A100 NVLINK, RTX 5090, RTX PRO6000 SE at 0; MON1 has A100 PCIe 8 (+4 spot), L40 30, RTX 4090 2, RTX A6000 1; OSL1 and TXS1 all 0

Console deploy page is viewable without login and shows in-stock SKUs with prices and a cost breakdown (running cost, stopped cost, deployable stock)

Docs FAQ gives 'latest on-demand GPU pricing (as of December 2024)' with 'SKY' H100/H200 rates that differ from the pricing page

Pricing-page FAQ says no spot instances, while console and availability list 'A100 80GB PCIE Spot'

Console menu shows 'Marketplace NEW' (rent a GPU / host your GPU) behind login

Reserve page offers 'Reserve your H200s' and 'Join the GB200 waitlist'

Changelog (July 2025): 'Full integration with Hyperstack API for private environment creation and management'

Home claims 9 data centers and 18,000+ GPUs; status page lists only MON1 and OSL1

Pages read

Alternatives

Providers people compare with it

GeoddOperator · 8 GPU types · H100 $3.00not scoredCompute CheapMarketplace · 4 GPU typesnot scoredThunder ComputeOperator · 4 GPU typesTrust 75HyperstackOperator · 10 GPU types · H100 $2.50Trust 81Impossible CloudOperator · 7 GPU typesnot scoredAceCloudOperator · 12 GPU typesTrust 73

Questions about Oblivus

What GPUs does Oblivus offer?

Oblivus has 9 GPU types with a fresh offer today, for example H100 from $3.50, A100 80GB from $1.47, RTX 4090 from $0.64, L40 from $1.05 per GPU-hour on-demand. The full list with prices and market medians is on this page.

How does Oblivus bill?

Billing granularity: per minute. Stopped instance: Storage and the assigned IPv4 only (NVMe 0.000154 USD/GB-hour + IP 0.0055 USD/h; console shows 0.0209 USD/h stopped for a 100 GB root disk). The IP can be detached while stopped to avoid the IP charge.

How trustworthy is Oblivus?

Its Trust Score is 82/100 (v0.1), read from its own documents. Strong on infrastructure control, pricing & billing transparency, reliability signals. Weaker on security & compliance. The score measures transparency and published guarantees, not customer satisfaction; every criterion and its source is listed above.

Where are Oblivus's data centres?

7 named regions in 2 countries: Canada, Norway.

How do you pay Oblivus, and are there credits?

Payment by card, invoice, crypto. Free credits: No free trial; occasional promo codes; referred sign-ups get 2% cashback (up to 25,000 USD); no formal academic credit program. Identity check: ID on request.