Marketplace · Santa Clara, CA
CloudRift
US software company (CloudRift Inc., Delaware) selling a GPU orchestration platform to datacenter operators and running a public GPU rental marketplace on partner datacenters, plus pay-per-token LLM inference.
GPU types10with a current offer
Current offers233233 readings on file · last 45 days
H100 from—per GPU-hour, on-demand
Entry price$0.25cheapest GPU-hour listed
Countries817 named regions
Trust Score v0.161/100
- Company & track record73
- Infrastructure control60
- Pricing & billing transparency85
- Reliability signals55
- Security & compliance27
- Customer signal—
Scored Sep 29, 2026 from the provider's own pages. Dimensions not yet measured are excluded from the denominator. Criteria ↓
Trust Score · 5 of 6 dimensions measuredThe 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 published | not disclosedcloudrift.ai · Sep 29, 2026 | 0 / 4 |
| Headquarters published | CloudRift Inc.cloudrift.ai · Sep 29, 2026 | 3 / 3 |
| Legal entity named | CloudRift Inc.cloudrift.ai · Sep 29, 2026 | 4 / 4 |
| Funding or ownership disclosed | Private; NVIDIA Inception member (home page). Funding not published.cloudrift.ai · Sep 29, 2026 | 4 / 4 |
Infrastructure control
weight 15%
60 / 100
| Capacity model published | third-party hostscloudrift.ai · Sep 29, 2026 | 4 / 10 |
| Regions published | 17 regions or countriescloudrift.ai · Sep 29, 2026 | 5 / 5 |
Pricing & billing transparency
weight 20%
85 / 100
| Public prices without an account | yes (read on the pricing page)cloudrift.ai · Sep 29, 2026 | 5 / 5 |
| Billing granularity published | per secondcloudrift.ai · Sep 29, 2026 | 4 / 4 |
| Egress policy published | included (pricing FAQ: 'no separate egress, ingress, or API-call charges')cloudrift.ai · Sep 29, 2026 | 4 / 4 |
| Stopped-instance billing published | Stopping a rental terminates it and ephemeral storage is lost; billing stops when Stop is pressed. Persistent volumes keep billing until deleted. An API 'pause' keeps resources allocated and billing continues. Reservations bill for the whole term regardless of instance state.cloudrift.ai · Sep 29, 2026 | 4 / 4 |
| Public IPv4 policy published | not published (public API shows public IP availability per datacenter; no price stated)cloudrift.ai · Sep 29, 2026 | 0 / 3 |
Reliability signals
weight 20%
55 / 100
| Status page | https://status.cloudrift.ai/cloudrift.ai · Sep 29, 2026 | 6 / 6 |
| SLA published | not disclosedcloudrift.ai · Sep 29, 2026 | 0 / 6 |
| SLA with a numeric uptime commitment | not disclosedcloudrift.ai · Sep 29, 2026 | 0 / 3 |
| Uptime history published | 99.99% shown as a headline figure on the GPU rentals page, without SLA terms; Terms state availability is not guaranteedcloudrift.ai · Sep 29, 2026 | 5 / 5 |
| Observed availability (30 days) | not measured yet | — |
Security & compliance
weight 15%
27 / 100
| SOC 2 | not published (security page: 'SOC 2 Type II, In Progress', monitored by Vanta; home page badge says 'SOC 2 Certified')cloudrift.ai · Sep 29, 2026 | 0 / 8 |
| ISO 27001 | no (some partner facilities described as ISO 27001, not CloudRift itself)cloudrift.ai · Sep 29, 2026 | 4 / 4 |
| HIPAA or GDPR DPA | not disclosedcloudrift.ai · Sep 29, 2026 | 0 / 3 |
Customer signal
weight 15%
not measured
| Aggregated public reviews | not 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 · 7 GPU types with a current offerWhat CloudRift 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.
| GPU | Cheapest here | Market median | vs market | Current offers | Modes | |
|---|---|---|---|---|---|---|
| A100 80GB | $1.05 | $1.99 | −47% | 8 | on-demand | A100 80GB prices |
| RTX PRO 6000 | $1.29 | $2.14 | −40% | 30 | on-demand | RTX PRO 6000 prices |
| L40S | $0.63 | $1.26 | −50% | 8 | on-demand | L40S prices |
| RTX 5090 | $0.60 | $0.69 | −13% | 42 | on-demand | RTX 5090 prices |
| RTX 4090 | $0.39 | $0.44 | −12% | 41 | on-demand | RTX 4090 prices |
| Tesla V100 16GB | $0.25 | $1.71 | −85% | 40 | on-demand | Tesla V100 16GB prices |
| MI350X | $4.00 | — | — | 8 | on-demand | MI350X prices |
All offers · 233 readings on fileEvery CloudRift 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.
| GPU | Configuration | Mode | Region | Stock | List | True cost | Read | |
|---|---|---|---|---|---|---|---|---|
| Tesla V100 16GBmarketplace flooroutlier · suspended | 1× Tesla V100 16GB12 vCPU · 52 GB RAM · 400 GB disk | On-demand | usny01_a01 | In stock | $0.25 | $0.25 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 2× Tesla V100 16GB24 vCPU · 104 GB RAM · 800 GB disk | On-demand | usny01_a01 | In stock | $0.25 | $0.25 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 3× Tesla V100 16GB36 vCPU · 156 GB RAM · 1,200 GB disk | On-demand | usny01_a01 | In stock | $0.25 | $0.25 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 6× Tesla V100 16GB72 vCPU · 312 GB RAM · 2,400 GB disk | On-demand | usny01_a01 | In stock | $0.25 | $0.25 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 8× Tesla V100 16GB96 vCPU · 416 GB RAM · 3,200 GB disk | On-demand | usny01_a01 | Unavailable | $0.25 | $0.25 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 4× Tesla V100 16GB48 vCPU · 208 GB RAM · 1,600 GB disk | On-demand | usny01_a01 | In stock | $0.25 | $0.25 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 5× Tesla V100 16GB60 vCPU · 260 GB RAM · 2,000 GB disk | On-demand | usny01_a01 | In stock | $0.25 | $0.25 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 7× Tesla V100 16GB84 vCPU · 364 GB RAM · 2,800 GB disk | On-demand | usny01_a01 | In stock | $0.25 | $0.25 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 1× Tesla V100 16GB8 vCPU · 85 GB RAM · 800 GB disk | On-demand | ustx1a_a01 | In stock | $0.28 | $0.28 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 1× Tesla V100 16GB8 vCPU · 85 GB RAM · 400 GB disk | On-demand | ustx1a_a01 | In stock | $0.28 | $0.28 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 13× Tesla V100 16GB130 vCPU · 1,105 GB RAM · 10,400 GB disk | On-demand | ustx1a_a01 | Unavailable | $0.29 | $0.29 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 14× Tesla V100 16GB140 vCPU · 1,190 GB RAM · 11,200 GB disk | On-demand | ustx1a_a01 | Unavailable | $0.29 | $0.29 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 15× Tesla V100 16GB150 vCPU · 1,275 GB RAM · 12,000 GB disk | On-demand | ustx1a_a01 | Unavailable | $0.29 | $0.29 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 16× Tesla V100 16GB160 vCPU · 1,360 GB RAM · 12,800 GB disk | On-demand | ustx1a_a01 | Unavailable | $0.29 | $0.29 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 2× Tesla V100 16GB20 vCPU · 170 GB RAM · 1,600 GB disk | On-demand | ustx1a_a01 | In stock | $0.29 | $0.29 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 3× Tesla V100 16GB30 vCPU · 255 GB RAM · 2,400 GB disk | On-demand | ustx1a_a01 | In stock | $0.29 | $0.29 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 4× Tesla V100 16GB40 vCPU · 340 GB RAM · 3,200 GB disk | On-demand | ustx1a_a01 | In stock | $0.29 | $0.29 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 5× Tesla V100 16GB50 vCPU · 425 GB RAM · 4,000 GB disk | On-demand | ustx1a_a01 | In stock | $0.29 | $0.29 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 6× Tesla V100 16GB60 vCPU · 510 GB RAM · 4,800 GB disk | On-demand | ustx1a_a01 | In stock | $0.29 | $0.29 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 7× Tesla V100 16GB70 vCPU · 595 GB RAM · 5,600 GB disk | On-demand | ustx1a_a01 | In stock | $0.29 | $0.29 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 8× Tesla V100 16GB80 vCPU · 680 GB RAM · 6,400 GB disk | On-demand | ustx1a_a01 | In stock | $0.29 | $0.29 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 9× Tesla V100 16GB90 vCPU · 765 GB RAM · 7,200 GB disk | On-demand | ustx1a_a01 | In stock | $0.29 | $0.29 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 10× Tesla V100 16GB100 vCPU · 850 GB RAM · 8,000 GB disk | On-demand | ustx1a_a01 | In stock | $0.29 | $0.29 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 11× Tesla V100 16GB110 vCPU · 935 GB RAM · 8,800 GB disk | On-demand | ustx1a_a01 | In stock | $0.29 | $0.29 | 21:10 UTCfresh | Receipt |
| Tesla V100 16GB | 12× Tesla V100 16GB120 vCPU · 1,020 GB RAM · 9,600 GB disk | On-demand | ustx1a_a01 | Unavailable | $0.29 | $0.29 | 21:10 UTCfresh | Receipt |
177 fresh offers · the 25 cheapest shown here, all of them in the comparison tool · prices per GPU-hour · true cost includes the add-ons the provider publishesOpen in the comparison tool →
True cost & billing · read Sep 29, 2026What 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 second
- Minimum charge
- None: no minimum duration and no minimum spend for on-demand (pricing FAQ).
- Stopped instance
- Stopping a rental terminates it and ephemeral storage is lost; billing stops when Stop is pressed. Persistent volumes keep billing until deleted. An API 'pause' keeps resources allocated and billing continues. Reservations bill for the whole term regardless of instance state.
- Egress
- included (pricing FAQ: 'no separate egress, ingress, or API-call charges')
- Public IPv4
- not published (public API shows public IP availability per datacenter; no price stated)
- Billing model
- prepaid credits
- Minimum deposit
- not published; Stripe deposits capped at 2,500 USD, larger amounts by invoice via [email protected]
- Free credits
- AI Grant: 100 to 1,000 USD in GPU credits on application, reviewed every two weeks
Account
- Payment
- card · invoice
- Sign-up
- email, password (or Google / GitHub sign-in)
- Identity check
- not published
- Support
- discord, email, contact form, phone
- Enterprise support
- not published
- SLA
- not published
- Status page
- status.cloudrift.ai ↗
- Published uptime
- 99.99% shown as a headline figure on the GPU rentals page, without SLA terms; Terms state availability is not guaranteed
Infrastructure
- Regions
17 named regions in 8 countries
ustx1a_a01 (Cato) (United States) · usny01_a01 (Cato) (United States) · us-east-nc-nr-1 / nr-2 / nr-ral (NeuralRack) (United States) · us-east-fl-nr-tpa (NeuralRack) (United States) · us-central-tx-ti-1 (TX1) (United States) · us-east-mn-ad-1 (AdvancedDatum · not listed on marketplace) (United States) · na-north-ca-mo-1 / mo-2 (1Legion) (Canada) · eu-west-uk-lo-1 / lo-2 / lo-3 (UK1) (United Kingdom) · eu-central-it-gv-1 (GraficaVeneta) (Italy) · eu-west-fr-az-1 (CloudRift · not listed on marketplace) (France) · ap-east-tw-kn-1 / kn-2 (KonstTech) (Taiwan) · ap_northeast_kr_se_1 (IRONGREY) (South Korea) · ca-central-kz-al-1 · ca-central-hc-al-1 / al-2 (Kazteleport · HyperCloud; not listed on marketplace) (Kazakhstan)- Countries
- United States, Canada, United Kingdom, Italy, France, Taiwan, South Korea, Kazakhstan
- Platform
- api, cli, serverless, jupyter, ssh, custom_docker_image, persistent_volumes, bare_metal · SDK : python (pip install cloudrift), rust API · Terraform : — · Kubernetes : no · interconnexion : NVLink 3.0 on A100, NVLink 2.0 on V100 (GPU pages) · no InfiniBand published · clusters : Multi-GPU nodes up to 8 GPUs (V100 SXM3 up to 16, RTX 4090 up to 10) · no multi-node cluster product published
- Storage
- ephemeral · included · Local storage per GPU as listed in the pricing table (396 GB to 2000 GB per GPU).volume · None · Persistent volumes billed independently of instances; price not published; only on some partner datacenters; deleted when the account balance is depleted.
- Migration out
- Workloads are standard KVM VMs or Docker containers with SSH/root access; data can be copied out over the network without egress fees. Persistent volumes are block devices that can be reattached to new instances but must be exported manually; no snapshot or image export is documented.
Security & compliance
- SOC 2
- not published (security page: 'SOC 2 Type II, In Progress', monitored by Vanta; home page badge says 'SOC 2 Certified')
- ISO 27001
- no (some partner facilities described as ISO 27001, not CloudRift itself)
- HIPAA
- not published
- GDPR / DPA
- not published
- Other
- Vanta trust report on request; bug bounty program
- Legal entity
- CloudRift Inc.
- Funding
- Private; NVIDIA Inception member (home page). Funding not published.
GPUBank market study, read on the provider's own pages · 26 pages read · facts marked "not published" were looked for and not foundPages read ↓
From the study · assessmentCloudRift, in short
What stood out when we read the documents: strengths, weaknesses, who it suits, and what we could not find.
Strengths
- Low published on-demand rates on older and consumer GPUs: V100 16GB $0.25, RTX 4090 $0.39, A100 SXM4 80GB $1.05 per GPU-hour
- Per-second billing, billing deferred until the VM signals readiness, no minimum spend, no egress or ingress charges
- VMs with IOMMU GPU passthrough, Docker containers and bare metal from one console, CLI and REST API (OpenAPI 3.1, MCP endpoint)
- Reservations of 1 week, 1 month and 3 months selectable in the console; running instances convertible to reservations
- AMD MI350X (288 GB) listed, with ROCm 7.2 VM images
- Pay-per-token OpenAI-compatible inference API on the same account
Weaknesses
- H100, H200, L40S and RTX PRO 6000 have no public on-demand price on the pricing page ('Upon request'); H100/H200 are not in the console
- Most instance types showed 0 available nodes in the public API at check time; only V100 types had stock
- Prices differ between pages (RTX 5090 $0.60 vs $0.65, MI350X $4.00 vs $3.65, V100 32GB $0.28 vs $0.29)
- Persistent volumes only in some datacenters, unpriced publicly, and deleted when the balance reaches zero
- No crypto payment; Stripe deposits capped at 2,500 USD
- All sales final, no refunds (Terms, section 6); no SLA; SOC 2 Type II still in progress
- Capacity sits in third-party partner datacenters; no spot tier
Ideal for
- Budget fine-tuning and batch inference on V100, RTX 4090 and A100 80GB
- Teams wanting VM or container GPU rentals with per-second billing and no egress fees
- Datacenter operators wanting white-label GPU orchestration software
Not published
- 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)
- Identity verification / KYC policy not published
- sign-up form sits behind the console and was not inspected
- 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
Observations
Pricing page shows a table of 11 GPU models (V100 SXM2 to MI350X) with On-Demand, 1 Month and 3 Month $/hr, VRAM, local storage and max GPUs, visible without an account
L40S, RTX PRO 6000, RTX PRO 6000 Max-Q, H100 and H200 show 'Upon request' in the on-demand column
Footnote under the table: 'Prices can fluctuate. Availability varies by demand and region.'
Interactive calculator: GPU model, reservation period (On-Demand, 1 Week 5% off, 1 Month 10% off, 3 Months 15% off), GPU count, hours per month (default 730); outputs total cost and effective rate per GPU
1-year reservations are 'contact us'
The calculator dropdown lists L40S at $0.57/hr, RTX PRO 6000 at $1.16/hr, H100 at $1.70/hr and H200 at $2.50/hr, i.e. their 1-month reserved rates, next to on-demand rates of the other models
Home page GPU cards: RTX 4090 $0.39 (ap-east-tw-kn-1, TW), A100 80GB $1.05 and V100 32GB $0.28 (ustx1a_a01, USA), RTX 5090 $0.60 'Out of stock', RTX PRO 6000 $1.16 'Reserved only', MI350X $3.65 'Out of stock'
Site is primarily marketed as a sovereign AI platform for datacenter operators; GPU rental is a secondary menu item. Footer shows an 'All Systems Operational' status badge
Pages read
- cloudrift.ai/pricing — GPU price table, calculator, reservation discounts, billing FAQ, egress
- cloudrift.ai/ — site observations, SOC 2 claim, GPU cards
- cloudrift.ai/gpu-rentals — 99.99% uptime headline, per-second billing, 8 locations, VM/container, Tier 3 facilities, price FAQ
- cloudrift.ai/rtx5090 — alternative RTX 5090 price quotes
- cloudrift.ai/mi350x — alternative MI350X price quotes
- cloudrift.ai/v100 — V100 prices, NVLink 2.0
- cloudrift.ai/a100 — A100 per-GPU resources, NVLink 3.0
- cloudrift.ai/about — founders, deployment countries
- cloudrift.ai/contact — legal entity, address, contact channels
- cloudrift.ai/security — SOC 2 Type II in progress, Vanta
- cloudrift.ai/terms — payment methods, no refunds, no availability guarantee
- cloudrift.ai/ai-grant — free credits 100 to 1,000 USD
- cloudrift.ai/for-enterprise — bare metal, MIG, platform scope
- docs.cloudrift.ai/features/volumes — volume behaviour, deletion at zero balance
- docs.cloudrift.ai/features/reservations — reservation conversion, Stripe 2,500 USD limit
- docs.cloudrift.ai/troubleshooting/volumes-and-reservations — stop = terminate, volume billing after termination
- docs.cloudrift.ai/troubleshooting/setup-and-general — billing stops at Stop, Discord support, Stripe billing
- docs.cloudrift.ai/extras/rest-api — API capabilities, MCP endpoint
- docs.cloudrift.ai/extras/python-api — Python SDK
- docs.cloudrift.ai/setup/cli — CLI, login methods
- docs.cloudrift.ai/cli-interface/instance-management — marketplace renting via CLI
- docs.cloudrift.ai/features/amd-gpu — MI300X/MI350X, ROCm 7.2
- docs.cloudrift.ai/changelog/2026 — billing deferred until VM ready, pause keeps billing
- api.cloudrift.ai/api/v1/instance-types/list — per-configuration prices, stock, datacenters
- api.cloudrift.ai/api/v1/providers/list — partner datacenters and countries
- status.cloudrift.ai/ — status page
Questions about CloudRift
What GPUs does CloudRift offer?
CloudRift has 10 GPU types with a fresh offer today, for example A100 80GB from $1.05, RTX PRO 6000 from $1.29, L40S from $0.63, RTX 5090 from $0.60, RTX 4090 from $0.39, Tesla V100 16GB from $0.25 per GPU-hour on-demand. The full list with prices and market medians is on this page.
How does CloudRift bill?
Billing granularity: per second. Minimum charge: None: no minimum duration and no minimum spend for on-demand (pricing FAQ). Stopped instance: Stopping a rental terminates it and ephemeral storage is lost; billing stops when Stop is pressed. Persistent volumes keep billing until deleted. An API 'pause' keeps resources allocated and billing continues. Reservations bill for the whole term regardless of instance state.
How trustworthy is CloudRift?
Its Trust Score is 61/100 (v0.1), read from its own documents. Weaker on infrastructure control, reliability signals, security & compliance. The score measures transparency and published guarantees, not customer satisfaction; every criterion and its source is listed above.
Where are CloudRift's data centres?
17 named regions in 8 countries: United States, Canada, United Kingdom, Italy, France, Taiwan, South Korea, Kazakhstan.
How do you pay CloudRift, and are there credits?
Payment by card, invoice. Free credits: AI Grant: 100 to 1,000 USD in GPU credits on application, reviewed every two weeks.