Marketplace · Lewes, DE
TensorDock
US GPU marketplace (TensorDock.com, Inc., Delaware) renting KVM virtual machines on vetted third-party hosts and partner data centers, with host-set hourly GPU prices, a la carte CPU/RAM/storage, spot bidding and a prepaid balance.
GPU types12with a current offer
Current offers5757 readings on file · last 45 days
H100 from$2.25per GPU-hour, on-demand
Entry price$0.10cheapest GPU-hour listed
Countries48 named regions
Trust Score v0.153/100
- Company & track record73
- Infrastructure control60
- Pricing & billing transparency100
- Reliability signals25
- Security & compliance0
- 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 disclosedtensordock.com · Sep 29, 2026 | 0 / 4 |
| Headquarters published | TensorDock.com, Inctensordock.com · Sep 29, 2026 | 3 / 3 |
| Legal entity named | TensorDock.com, Inctensordock.com · Sep 29, 2026 | 4 / 4 |
| Funding or ownership disclosed | Private; all-remote team (about page). Funding not published.tensordock.com · Sep 29, 2026 | 4 / 4 |
Infrastructure control
weight 15%
60 / 100
| Capacity model published | third-party hoststensordock.com · Sep 29, 2026 | 4 / 10 |
| Regions published | 8 regions or countriestensordock.com · Sep 29, 2026 | 5 / 5 |
Pricing & billing transparency
weight 20%
100 / 100
| Public prices without an account | yes (read on the pricing page)tensordock.com · Sep 29, 2026 | 5 / 5 |
| Billing granularity published | per secondtensordock.com · Sep 29, 2026 | 4 / 4 |
| Egress policy published | included ('no ingress/egress fees'; minimum 1 Gbps included)tensordock.com · Sep 29, 2026 | 4 / 4 |
| Stopped-instance billing published | Storage only at the host's storage rate (deploy page estimate: 100 GB stopped = $0.0050/hour); GPUs are released ('stop and unreserve') and restart depends on availability. Core Compute locations use network storage so a stopped VM can restart on any server with that GPU in the same location.tensordock.com · Sep 29, 2026 | 4 / 4 |
| Public IPv4 policy published | Port forwarding by default; dedicated IP where the location supports it (Core Compute: static IP per VM). Dedicated IP price not published.tensordock.com · Sep 29, 2026 | 3 / 3 |
Reliability signals
weight 20%
25 / 100
| Status page | not disclosedtensordock.com · Sep 29, 2026 | 0 / 6 |
| SLA published | not disclosedtensordock.com · Sep 29, 2026 | 0 / 6 |
| SLA with a numeric uptime commitment | not disclosedtensordock.com · Sep 29, 2026 | 0 / 3 |
| Uptime history published | Per-hostnode uptime_percentage field in API v2tensordock.com · Sep 29, 2026 | 5 / 5 |
| Observed availability (30 days) | not measured yet | — |
Security & compliance
weight 15%
0 / 100
| SOC 2 | not disclosedtensordock.com · Sep 29, 2026 | 0 / 8 |
| ISO 27001 | not disclosedtensordock.com · Sep 29, 2026 | 0 / 4 |
| HIPAA or GDPR DPA | not disclosedtensordock.com · 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 →
Market report · Oct 25, 2026 · Specialised GPU clouds, hourly rental, pods and inference · #45 of the 200 to audit · prices read on the provider's pageWhat the market report notes about TensorDock
Our internal report lists 248 actors with their zone, segment, status and the prices it could read, dated Oct 25, 2026. Its notes are reproduced as written, in French, with their sources.
- Listed as
- TensorDock
Prices by GPU · 12 GPU types with a current offerWhat TensorDock 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 | |
|---|---|---|---|---|---|---|
| H100 | $2.25 | $3.17 | −29% | 10 | reservedspoton-demand | H100 prices |
| A100 80GB | $1.50 | $1.99 | −25% | 3 | spoton-demand | A100 80GB prices |
| RTX A6000 | $0.45 | $0.54 | −17% | 1 | on-demand | RTX A6000 prices |
| RTX 4090 | $0.35 | $0.44 | −20% | 3 | spoton-demand | RTX 4090 prices |
| RTX 6000 Ada | $0.75 | $0.79 | −5% | 1 | on-demand | RTX 6000 Ada prices |
| Tesla V100 32GB | $0.42 | $2.48 | −83% | 1 | on-demand | Tesla V100 32GB prices |
| L40 | $0.95 | $0.89 | +7% | 2 | on-demand | L40 prices |
| Tesla V100 16GB | $0.17 | $1.71 | −90% | 1 | on-demand | Tesla V100 16GB prices |
| RTX 3090 | $0.20 | $0.19 | +5% | 1 | on-demand | RTX 3090 prices |
| RTX A4000 | $0.10 | $0.19 | −46% | 1 | on-demand | RTX A4000 prices |
| RTX A5000 | $0.27 | $0.27 | +1% | 1 | on-demand | RTX A5000 prices |
| Quadro RTX 8000 | $0.32 | — | — | 1 | on-demand | Quadro RTX 8000 prices |
All offers · 57 readings on fileEvery TensorDock 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 | |
|---|---|---|---|---|---|---|---|---|
| RTX A4000aggregator reading | RTX A4000 | On-demand | — | Stock unknown | $0.10 | $0.10 | Oct 25, 2026older reading | Receipt |
| Tesla V100 16GBSXMmarketplace flooroutlier · suspended | Tesla V100 16GB SXM | On-demand | TensorDock marketplace (host-set prices) | Stock unknown | $0.17 | $0.17 | Sep 29, 2026older reading | Receipt |
| RTX 4090marketplace floor | RTX 4090 | Spot | US | Stock unknown | $0.20 | $0.20 | Sep 29, 2026older reading | Receipt |
| RTX 4090 | RTX 4090 | Spot | US | Stock unknown | $0.20 | $0.20 | Sep 29, 2026older reading | Receipt |
| RTX 3090aggregator reading | RTX 3090 | On-demand | — | Stock unknown | $0.20 | $0.20 | Oct 25, 2026older reading | Receipt |
| RTX A5000marketplace floor | RTX A5000 | On-demand | TensorDock marketplace (host-set prices) | Stock unknown | $0.27 | $0.27 | Sep 29, 2026older reading | Receipt |
| Quadro RTX 8000marketplace floor | Quadro RTX 8000 | On-demand | TensorDock marketplace (host-set prices) | Stock unknown | $0.32 | $0.32 | Sep 29, 2026older reading | Receipt |
| RTX 4090aggregator reading | RTX 4090 | On-demand | — | Stock unknown | $0.35 | $0.35 | Oct 25, 2026older reading | Receipt |
| Tesla V100 32GBmarketplace flooroutlier · suspended | Tesla V100 32GB | On-demand | TensorDock marketplace (host-set prices) | Stock unknown | $0.42 | $0.42 | Sep 29, 2026older reading | Receipt |
| RTX A6000aggregator reading | RTX A6000 | On-demand | — | Stock unknown | $0.45 | $0.45 | Oct 25, 2026older reading | Receipt |
| A100 80GBSXMmarketplace floor | 1× A100 80GB SXM | Spot | — | Stock unknown | $0.67 | $0.67 | Oct 16, 2026older reading | Receipt |
| RTX 6000 Adaaggregator reading | RTX 6000 Ada | On-demand | — | Stock unknown | $0.75 | $0.75 | Oct 25, 2026older reading | Receipt |
| L40marketplace floor | L40 | On-demand | TensorDock marketplace (host-set prices) | Stock unknown | $0.95 | $0.95 | Sep 29, 2026older reading | Receipt |
| L40 | L40 | On-demand | TensorDock marketplace (host-set prices) | Stock unknown | $1.07 | $1.07 | Sep 29, 2026older reading | Receipt |
| H100SXMmarketplace floor | H100 SXM | Reserved | US | Stock unknown | $1.50 | $1.50 | Sep 29, 2026older reading | Receipt |
| H100SXM3 years | H100 SXM | Reserved | US | Stock unknown | $1.50 | $1.50 | Sep 29, 2026older reading | Receipt |
| H100SXM3 years | H100 SXM | Reserved | US | Stock unknown | $1.50 | $1.50 | Sep 29, 2026older reading | Receipt |
| A100 80GBPCIEaggregator reading | A100 80GB PCIE | On-demand | — | Stock unknown | $1.50 | $1.50 | Oct 25, 2026older reading | Receipt |
| A100 80GBSXMaggregator reading | A100 80GB SXM | On-demand | — | Stock unknown | $1.80 | $1.80 | Oct 25, 2026older reading | Receipt |
| H100SXM1 year (Annual) | H100 SXM | Reserved | US | Stock unknown | $1.90 | $1.90 | Sep 29, 2026older reading | Receipt |
| H100SXM1 year (Annual) | H100 SXM | Reserved | US | Stock unknown | $1.90 | $1.90 | Sep 29, 2026older reading | Receipt |
| H100SXMmarketplace floor | H100 SXM | Spot | US | Stock unknown | $1.91 | $1.91 | Sep 29, 2026older reading | Receipt |
| H100SXM | H100 SXM | Spot | US | Stock unknown | $1.91 | $1.91 | Sep 29, 2026older reading | Receipt |
| H100SXM1 month (Monthly) | H100 SXM | Reserved | US | Stock unknown | $2.00 | $2.00 | Sep 29, 2026older reading | Receipt |
| H100SXM1 month (Monthly) | H100 SXM | Reserved | US | Stock unknown | $2.00 | $2.00 | Sep 29, 2026older reading | Receipt |
26 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
- No minimum per deployment ('Billed per second, no minimum' on deploy page). API instance creation requires a minimum balance of $1 and minimum 100 GB storage.
- Stopped instance
- Storage only at the host's storage rate (deploy page estimate: 100 GB stopped = $0.0050/hour); GPUs are released ('stop and unreserve') and restart depends on availability. Core Compute locations use network storage so a stopped VM can restart on any server with that GPU in the same location.
- Egress
- included ('no ingress/egress fees'; minimum 1 Gbps included)
- Public IPv4
- Port forwarding by default; dedicated IP where the location supports it (Core Compute: static IP per VM). Dedicated IP price not published.
- Billing model
- prepaid credits
- Minimum deposit
- 5 USD ('Start with as little as $5')
- Free credits
- not published
Account
- Payment
- card · crypto (yes (implied): security page states 'We conduct KYC checks for those paying via other means, such as cryptocurrency'; coins and payment rail not published)
- Sign-up
- email, password, card (3D Secure attempted on every card transaction), KYC for non-card payments such as cryptocurrency, VAT/GST business ID for customers in VAT/GST countries
- Identity check
- card verification
- Support
- email, live chat, discord
- Enterprise support
- not published
- SLA
- No customer uptime SLA. Hosts held to a 99.99% uptime standard; downtime credits 10x hourly rate (Top Hosts) or 5x (Standard Hosts), min $5, max 100% of previous month's spend; data loss credit 40x hourly rate.
- Status page
- not published
- Published uptime
- Per-hostnode uptime_percentage field in API v2
Infrastructure
- Regions
8 named regions in 4 countries
Dallas (Evoque · Voltage Park partner) (United States) · Hillsboro (Flexential) (United States) · Manassas (Core Compute) (United States) · Joplin (United States) · Devon (United Kingdom) · Prague (Czechia) · Germany (city not stated) (Germany)- Countries
- United States, United Kingdom, Czech Republic, Germany
- Platform
- api, jupyter, ssh, persistent_volumes, spot_or_interruptible, bare_metal · SDK : — · Terraform : — · Kubernetes : no · interconnexion : — · clusters : 8x H100 bare metal hostnodes on monthly/annual/3-year terms · 8x A100/4090 hostnode reservations via sales
- Storage
- volume · 0.00005 USD/GB-hour (0.036 USD/GB-month at 720 h) · Block NVMe SSD, billed per second, set per host. API requires 100 GB minimum per VM (marketing table says 20 GB minimum). Storage can be increased, never decreased.network · None · Network storage available in Core Compute locations; separate price not published.
- Migration out
- VMs are standard KVM images; users can request an export of the VM disk image (GPU pages) or copy data out with SCP/rclone (docs). No egress fees. Storage cannot be shrunk once increased.
Security & compliance
- SOC 2
- not published
- ISO 27001
- not published
- HIPAA
- not published
- GDPR / DPA
- not published
- Other
- No company-level certification published; security page states partner data centers hold certifications ranging from SOC 2, PCI DSS, ISO 27001 and HIPAA (per data center, on request).
- Legal entity
- TensorDock.com, Inc
- Funding
- Private; all-remote team (about page). Funding not published.
GPUBank market study, read on the provider's own pages · 25 pages read · facts marked "not published" were looked for and not foundPages read ↓
From the study · assessmentTensorDock, in short
What stood out when we read the documents: strengths, weaknesses, who it suits, and what we could not find.
Strengths
- Full KVM virtual machines with root access and dedicated GPU passthrough, Windows 10 supported (home page)
- Per-second billing with a la carte vCPU ($0.003/h), RAM ($0.002/GB-h) and NVMe ($0.00005/GB-h) (cloud-gpus page)
- No ingress/egress fees (cloud-gpus page)
- Stopped VMs billed for storage only; deploy page shows 100 GB stopped at $0.0050/hour
- H100 SXM reserved per-GPU rates published: $2.00 monthly, $1.90 annual, $1.50 3-year (gpu-h100 page)
- Spot bidding, host minimum bids 'typically around 50% of the on-demand price' (spot docs)
- Published downtime credits of 5x to 10x the hourly rate and 40x for data loss (docs)
- REST API v2 with location and hostnode discovery, cloud-init and instance resizing (dashboard API docs)
Weaknesses
- No live price list without an account: the deploy page and the public locations/GPU endpoints return no GPU data when signed out
- Public price tables are dated July 2024 and conflict between pages (H100 spot $1.91 vs $1.30; A100 $1.42 / $1.50 / $1.63)
- Headline GPU prices exclude CPU, RAM and storage
- Prices set by independent hosts and vary by host; 'Standard Hosts' may be in residential or business facilities with little redundancy
- Servers are deleted automatically when the balance reaches $0 (home page FAQ)
- Cannot serve consumers in VAT/GST countries without a VAT/GST ID (taxes page)
- No serverless, managed Kubernetes or container product documented; no company-level compliance certification
Ideal for
- Cost-sensitive users who want full VMs (including Windows and cloud gaming) on consumer or older data-center GPUs
- API integrators who want to choose exact hosts and pay a la carte for CPU/RAM
- Teams ready to sign multi-month H100 reservations
Not published
- Live per-host GPU prices and the full current catalogue: dashboard and API require login
- public endpoints return empty data
- Which crypto coins or processor are accepted, and the KYC process for crypto payments
- Dedicated IPv4 price and network storage price
- Whether PayPal or wire/invoice are accepted
- Founding year and funding
- Any CLI, SDK or Terraform provider (only REST API v2 documented
- an older Postman v0 collection is linked from the site)
- Whether the 2024-dated price tables are still current
Observations
https://www.tensordock.com/pricing returns 404; the only public GPU price table is on cloud-gpus.html, marked 'Last Updated: July 24, 2024. Please check the dashboard for live pricing and availability.'
A site-wide banner still reads 'March News: TensorDock launches a massive fleet of on-demand NVIDIA H100 SXMs at just $2.25/hr'
Home page claims '45 available GPU models'; GPU pages say '27 GPU types'; comparison pages say 44; only 14 models are named on public pages
Deploy page (dashboard.tensordock.com/deploy) is viewable signed out but shows 'Loading available GPUs' and 'Authentication Required'; its estimator shows $0.0490/h running and $0.0050/h stopped for 4 vCPU/16 GB/100 GB, 'Billed per second, no minimum, Monthly estimates based on 730 hours'
The advertised 'curl https://dashboard.tensordock.com/api/stock/list' returns 404; /api/v2/locations returns an empty list unauthenticated
Home page counters show '97+ Locations', '703+ GPUs in stock', '29,703+ GPUs through partners'
Comparison pages (vs AWS, Azure, GCP, CoreWeave, Lambda, Paperspace, RunPod, Vast) are dated 2024 and carry a disclaimer that the data may be inaccurate
RunPod comparison page states balance refunds on request minus a 5% fee
Pages read
- tensordock.com/ — Headline prices, billing FAQ, $5 start, location counters, 45 models claim
- tensordock.com/cloud-gpus.html — On-demand price table, a la carte resource prices, no egress fees
- tensordock.com/gpu-h100.html — H100 on-demand, spot, monthly/annual/3-year and bare metal prices, Dallas location
- tensordock.com/gpu-a100.html — A100 prices, Devon/Prague/Oregon locations
- tensordock.com/gpu-4090.html — 4090 on-demand and spot prices
- tensordock.com/benchmarks.html — Price labels for 11 GPUs incl. H100 PCIe, V100 32GB, Quadro 8000
- tensordock.com/comparison-runpod.html — Equivalent configuration prices, 5% refund fee, support channels
- tensordock.com/comparison-vast.html — GPU type count, 1 Gbps minimum
- tensordock.com/security.html — 3D Secure, KYC for crypto payments, compliance statements
- tensordock.com/about.html — 24/7 staff, same pricing for all, no referral program
- tensordock.com/host.html — Host requirements, 25% revenue-share fee
- tensordock.com/cloud-cpus.html — CPU locations in US, Europe, India
- dashboard.tensordock.com/deploy — Per-second billing, stopped cost estimate, login requirement
- dashboard.tensordock.com/api/docs/getting-started — Bearer auth, 100 req/min rate limit, base URL
- dashboard.tensordock.com/api/docs/fetch-hostnodes — API structure, max_count 8, per-host pricing fields
- dashboard.tensordock.com/api/docs/instance-creation — Minimum $1 balance, 100 GB storage minimum, OS images
- dashboard.tensordock.com/api/docs/instance-management — Stop/start/modify, storage cannot decrease
- dashboard.tensordock.com/api/v2/locations — Returns empty list when signed out
- docs.tensordock.com/virtual-machines/spot-instances — Bid mechanics, minimum bid about 50% of on-demand
- docs.tensordock.com/virtual-machines/introduction-to-core-compute-vms — Network storage, static IPs, Manassas location
- docs.tensordock.com/ — 100+ locations, Joplin and Germany, support email
- docs.tensordock.com/legal-information/legal-information — Legal entity and address
- docs.tensordock.com/legal-information/terms-of-service-tos — Account verification rights, USD pricing, prepayment
- docs.tensordock.com/legal-information/downtime-compensation — Credit multipliers
- docs.tensordock.com/legal-information/taxes-vat-gst — VAT/GST ID requirement
Questions about TensorDock
What GPUs does TensorDock offer?
TensorDock has 12 GPU types with a fresh offer today, for example H100 from $2.25, A100 80GB from $1.50, RTX A6000 from $0.45, RTX 4090 from $0.35, RTX 6000 Ada from $0.75, L40 from $0.95 per GPU-hour on-demand. The full list with prices and market medians is on this page.
How does TensorDock bill?
Billing granularity: per second. Minimum charge: No minimum per deployment ('Billed per second, no minimum' on deploy page). API instance creation requires a minimum balance of $1 and minimum 100 GB storage. Stopped instance: Storage only at the host's storage rate (deploy page estimate: 100 GB stopped = $0.0050/hour); GPUs are released ('stop and unreserve') and restart depends on availability. Core Compute locations use network storage so a stopped VM can restart on any server with that GPU in the same location.
How trustworthy is TensorDock?
Its Trust Score is 53/100 (v0.1), read from its own documents. Strong on pricing & billing transparency. 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 TensorDock's data centres?
8 named regions in 4 countries: United States, United Kingdom, Czech Republic, Germany.
How do you pay TensorDock, and are there credits?
Payment by card, crypto. No sign-up credit is published. Identity check: card verification.