Marketplace · Los Angeles, CA · founded 2016
Vast.ai
Rent cloud GPUs from anywhere. Save up to 90% vs AWS, Azure, GCP with a few clicks.
GPU types91with a current offer
Current offers7,6237,623 readings on file · last 45 days
H100 from$2.27per GPU-hour, on-demand
Entry price$0.03cheapest GPU-hour listed
Countries52regions not published
Trust Score v0.178/100
- Company & track record100
- Infrastructure control60
- Pricing & billing transparency100
- Reliability signals55
- Security & compliance73
- Customer signal—
Scored Sep 29, 2026 from the provider's own pages. Dimensions not yet measured are excluded from the denominator. Criteria ↓
Outbound link carries a commission. It does not affect this page.
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%
100 / 100
| Founding year published | 2016vast.ai · Sep 29, 2026 | 4 / 4 |
| Headquarters published | Vast.ai Inc.vast.ai · Sep 29, 2026 | 3 / 3 |
| Legal entity named | Vast.ai Inc.vast.ai · Sep 29, 2026 | 4 / 4 |
| Funding or ownership disclosed | Private; incorporated as a Delaware C corporation on 28 June 2016, launched September 2018; funding not published on its sitevast.ai · Sep 29, 2026 | 4 / 4 |
Infrastructure control
weight 15%
60 / 100
| Capacity model published | third-party hostsvast.ai · Sep 29, 2026 | 4 / 10 |
| Regions published | 52 regions or countriesvast.ai · Sep 29, 2026 | 5 / 5 |
Pricing & billing transparency
weight 20%
100 / 100
| Public prices without an account | yes (read on the pricing page)vast.ai · Sep 29, 2026 | 5 / 5 |
| Billing granularity published | per secondvast.ai · Sep 29, 2026 | 4 / 4 |
| Egress policy published | metered: per byte, upload and download, host-set; internet cost in the public search API ranged 0 to 68.27 USD/TB across sampled offers (median 2.67 USD/TB)vast.ai · Sep 29, 2026 | 4 / 4 |
| Stopped-instance billing published | GPU charge stops; container disk keeps being billed per second at the host's storage rate until the instance is destroyed; docs note stopped-instance storage rates can be higher than running ratesvast.ai · Sep 29, 2026 | 4 / 4 |
| Public IPv4 policy published | shared/NAT by default (internal ports mapped to random external ports on the host IP); some hosts list static IPsvast.ai · Sep 29, 2026 | 3 / 3 |
Reliability signals
weight 20%
55 / 100
| Status page | https://status.vast.ai/vast.ai · Sep 29, 2026 | 6 / 6 |
| SLA published | not disclosedvast.ai · Sep 29, 2026 | 0 / 6 |
| SLA with a numeric uptime commitment | not disclosedvast.ai · Sep 29, 2026 | 0 / 3 |
| Uptime history published | Status page on 2026-09-29: vast.ai 99.95%, console 99.75%, serverless 99.75% over the last 30 daysvast.ai · Sep 29, 2026 | 5 / 5 |
| Observed availability (30 days) | not measured yet | — |
Security & compliance
weight 15%
73 / 100
| SOC 2 | Type IIvast.ai · Sep 29, 2026 | 8 / 8 |
| ISO 27001 | not published for Vast.ai itself; Secure Cloud datacenter partners may hold ISO 27001vast.ai · Sep 29, 2026 | 0 / 4 |
| HIPAA or GDPR DPA | Supported on Secure Cloud tier; BAAs for qualifying customersvast.ai · Sep 29, 2026 | 3 / 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 · #37 of the 200 to audit · prices read on the provider's pageWhat the market report notes about Vast.ai
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
- Vast.ai
- Cross-checkfr
- Marketplace : H100 dès 1,68 $ (hôte non vérifié), spot dès 0,34 $ ; RTX 5090 plus fiable qu'ailleurs.
Prices by GPU · 82 GPU types with a current offerWhat Vast.ai 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 | $2.27 | $3.17 | −28% | 48 | on-demandspot | H100 prices |
| A100 80GB | $0.60 | $1.99 | −70% | 82 | on-demandspot | A100 80GB prices |
| RTX PRO 6000 | $0.83 | $2.14 | −61% | 380 | on-demandspot | RTX PRO 6000 prices |
| H200 | $2.76 | $4.76 | −42% | 30 | on-demandspot | H200 prices |
| L40S | $0.34 | $1.26 | −73% | 54 | on-demandspot | L40S prices |
| B200 | $7.50 | $7.50 | +0% | 8 | on-demandspot | B200 prices |
| B300 | $8.50 | $8.96 | −5% | 12 | on-demandspot | B300 prices |
| L4 | $0.10 | $0.80 | −87% | 38 | on-demandspot | L4 prices |
| RTX A6000 | $0.36 | $0.54 | −33% | 26 | on-demandspot | RTX A6000 prices |
| RTX 5090 | $0.40 | $0.69 | −42% | 1024 | on-demandspot | RTX 5090 prices |
| T4 | $0.04 | $0.51 | −92% | 36 | on-demandspot | T4 prices |
| RTX 4090 | $0.13 | $0.44 | −69% | 814 | on-demandspot | RTX 4090 prices |
| A100 40GB | $0.34 | $1.98 | −83% | 110 | on-demandspot | A100 40GB prices |
| RTX 6000 Ada | $0.49 | $0.79 | −37% | 32 | on-demandspot | RTX 6000 Ada prices |
| Tesla V100 32GB | $0.07 | $2.48 | −97% | 150 | on-demandspot | Tesla V100 32GB prices |
| L40 | $0.27 | $0.89 | −70% | 10 | on-demandspot | L40 prices |
| Tesla V100 16GB | $0.07 | $1.71 | −96% | 38 | on-demandspot | Tesla V100 16GB prices |
| RTX 3090 | $0.10 | $0.19 | −45% | 520 | on-demandspot | RTX 3090 prices |
| H100 NVL | $2.61 | $3.94 | −34% | 14 | on-demandspot | H100 NVL prices |
| H200 NVL | $2.67 | $3.99 | −33% | 94 | on-demandspot | H200 NVL prices |
| RTX A4000 | $0.07 | $0.19 | −61% | 139 | on-demandspot | RTX A4000 prices |
| RTX A5000 | $0.19 | $0.27 | −30% | 34 | on-demandspot | RTX A5000 prices |
| RTX 5060 Ti | $0.09 | $0.14 | −33% | 328 | on-demandspot | RTX 5060 Ti prices |
| A10 | $0.24 | $1.35 | −82% | 18 | on-demandspot | A10 prices |
| RTX 5070 Ti | $0.17 | $0.25 | −30% | 186 | on-demandspot | RTX 5070 Ti prices |
| RTX 4000 Ada | $0.17 | $0.27 | −35% | 22 | on-demandspot | RTX 4000 Ada prices |
| RTX 5080 | $0.19 | $0.27 | −30% | 270 | on-demandspot | RTX 5080 prices |
| RTX PRO 5000 | $0.48 | $0.93 | −48% | 98 | on-demandspot | RTX PRO 5000 prices |
| A40 | $0.48 | $0.85 | −43% | 8 | on-demandspot | A40 prices |
| RTX 3060 Ti | $0.05 | $0.07 | −16% | 58 | on-demandspot | RTX 3060 Ti prices |
| RTX 3080 | $0.08 | $0.13 | −34% | 174 | on-demandspot | RTX 3080 prices |
| RTX 5070 | $0.12 | $0.19 | −36% | 48 | on-demandspot | RTX 5070 prices |
| RTX 4070 | $0.12 | $0.14 | −14% | 32 | on-demandspot | RTX 4070 prices |
| RTX 4080 | $0.20 | $0.25 | −19% | 32 | on-demandspot | RTX 4080 prices |
| RTX 3060 | $0.03 | $0.05 | −36% | 382 | on-demandspot | RTX 3060 prices |
| RTX 2080 Ti | $0.07 | $0.10 | −28% | 74 | on-demandspot | RTX 2080 Ti prices |
| RTX 3070 | $0.08 | $0.08 | +1% | 100 | on-demandspot | RTX 3070 prices |
| RTX 4070 Super | $0.09 | $0.14 | −34% | 102 | on-demandspot | RTX 4070 Super prices |
| RTX 4060 Ti | $0.10 | $0.12 | −14% | 104 | on-demandspot | RTX 4060 Ti prices |
| RTX 4080 Super | $0.21 | $0.23 | −8% | 32 | on-demandspot | RTX 4080 Super prices |
| RTX 2070 Super | $0.06 | $0.07 | −13% | 18 | on-demandspot | RTX 2070 Super prices |
| RTX 5060 | $0.08 | $0.10 | −19% | 22 | on-demandspot | RTX 5060 prices |
| RTX 3080 Ti | $0.08 | $0.10 | −22% | 84 | on-demandspot | RTX 3080 Ti prices |
| RTX 4070 Ti | $0.15 | $0.15 | −1% | 18 | on-demandspot | RTX 4070 Ti prices |
| RTX 3070 Ti | $0.16 | $0.13 | +24% | 6 | on-demandspot | RTX 3070 Ti prices |
| RTX PRO 4000 | $0.18 | $0.27 | −32% | 44 | on-demandspot | RTX PRO 4000 prices |
| RTX 3090 Ti | $0.20 | $0.20 | −0% | 34 | on-demandspot | RTX 3090 Ti prices |
| RTX PRO 4500 | $0.35 | $0.72 | −51% | 4 | on-demandspot | RTX PRO 4500 prices |
| RTX A2000 | $0.03 | $0.05 | −35% | 22 | on-demandspot | RTX A2000 prices |
| Tesla P4 | $0.04 | $0.07 | −41% | 2 | on-demandspot | Tesla P4 prices |
| RTX 4060 | $0.05 | $0.09 | −37% | 12 | on-demandspot | RTX 4060 prices |
| RTX 2060 | $0.06 | $0.05 | +15% | 34 | on-demandspot | RTX 2060 prices |
| GTX 1080 Ti | $0.06 | $0.08 | −24% | 18 | on-demandspot | GTX 1080 Ti prices |
| Tesla P40 | $0.11 | $0.09 | +18% | 2 | on-demandspot | Tesla P40 prices |
| CMP 170HX | $0.27 | $0.48 | −44% | 88 | on-demandspot | CMP 170HX prices |
| RTX 5000 Ada | $2.12 | $1.36 | +56% | 4 | on-demandspot | RTX 5000 Ada prices |
| GTX 1660 Ti | $0.04 | $0.04 | −0% | 6 | on-demandspot | GTX 1660 Ti prices |
| Tesla P100 | $0.04 | $0.07 | −40% | 32 | on-demandspot | Tesla P100 prices |
| GTX 1070 Ti | $0.05 | $0.05 | −0% | 16 | on-demandspot | GTX 1070 Ti prices |
| RTX 2070 | $0.05 | $0.06 | −10% | 24 | on-demandspot | RTX 2070 prices |
| RTX 2060 Super | $0.07 | $0.06 | +4% | 20 | on-demandspot | RTX 2060 Super prices |
| RTX 2080 | $0.09 | $0.08 | +7% | 2 | on-demandspot | RTX 2080 prices |
| Quadro RTX 5000 | $0.10 | $0.18 | −44% | 4 | on-demandspot | Quadro RTX 5000 prices |
| Titan Xp | $0.04 | — | — | 36 | on-demandspot | Titan Xp prices |
| GTX 1060 | $0.04 | — | — | 6 | on-demandspot | GTX 1060 prices |
| GTX 1080 | $0.04 | — | — | 28 | on-demandspot | GTX 1080 prices |
| GTX 1070 | $0.05 | — | — | 8 | on-demandspot | GTX 1070 prices |
| Quadro P4000 | $0.07 | — | — | 8 | on-demandspot | Quadro P4000 prices |
| Quadro RTX 4000 | $0.11 | — | — | 2 | on-demandspot | Quadro RTX 4000 prices |
| RTX 6000 | $0.13 | — | — | 16 | on-demandspot | RTX 6000 prices |
| Titan RTX | $0.13 | — | — | 18 | on-demandspot | Titan RTX prices |
| Titan V | $0.20 | — | — | 8 | on-demandspot | Titan V prices |
| Quadro RTX 8000 | $0.25 | — | — | 20 | on-demandspot | Quadro RTX 8000 prices |
| RTX 4500 Ada | $0.37 | — | — | 4 | on-demandspot | RTX 4500 Ada prices |
| RTX 5880 Ada | $0.80 | — | — | 12 | on-demandspot | RTX 5880 Ada prices |
| Quadro P2000 | $0.03 | — | — | 6 | on-demandspot | Quadro P2000 prices |
| GTX 1660 | $0.03 | — | — | 28 | on-demandspot | GTX 1660 prices |
| GTX Titan X | $0.04 | — | — | 6 | on-demandspot | GTX Titan X prices |
| GTX 1650 | $0.05 | — | — | 4 | on-demandspot | GTX 1650 prices |
| GTX 1050 Ti | $0.28 | — | — | 2 | on-demandspot | GTX 1050 Ti prices |
| A800 80GB PCIe | $0.47 | — | — | 8 | on-demandspot | A800 80GB PCIe prices |
| A100X | $1.28 | — | — | 6 | on-demandspot | A100X prices |
All offers · 7,623 readings on fileEvery Vast.ai 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 | |
|---|---|---|---|---|---|---|---|---|
| GTX 1050 Timarketplace floor | 1× GTX 1050 Ti4 vCPU · 7.7 GB RAM · 20.2 GB disk | Spot | Vietnam | In stock | $0.01 | $0.01 | 21:05 UTCfresh | Receipt |
| GTX 1060marketplace floor | 1× GTX 10608 vCPU · 15.6 GB RAM · 748.1 GB disk | Spot | Canada | In stock | $0.01 | $0.01 | 21:05 UTCfresh | Receipt |
| Tesla V100 16GBmarketplace flooroutlier · suspended | 1× Tesla V100 16GB8 vCPU · 31.3 GB RAM · 8,163 GB disk | Spot | Saudi Arabia | In stock | $0.01 | $0.01 | 21:05 UTCfresh | Receipt |
| Quadro P2000marketplace floor | 1× Quadro P20005 vCPU · 6.8 GB RAM · 32.2 GB disk | Spot | US | In stock | $0.01 | $0.01 | 21:05 UTCfresh | Receipt |
| Quadro P2000 | 4× Quadro P200020 vCPU · 27.3 GB RAM · 128.8 GB disk | Spot | US | In stock | $0.01 | $0.01 | 21:05 UTCfresh | Receipt |
| Quadro P2000 | 2× Quadro P200010 vCPU · 13.7 GB RAM · 64.4 GB disk | Spot | US | In stock | $0.01 | $0.01 | 21:05 UTCfresh | Receipt |
| GTX 1060 | 1× GTX 106020 vCPU · 15.5 GB RAM · 138.3 GB disk | Spot | Vietnam | In stock | $0.02 | $0.02 | 21:07 UTCfresh | Receipt |
| GTX 1660marketplace floor | 1× GTX 166012 vCPU · 31.3 GB RAM · 645 GB disk | Spot | Austria | In stock | $0.02 | $0.02 | 21:06 UTCfresh | Receipt |
| GTX 1660 | 1× GTX 166012 vCPU · 31.3 GB RAM · 724 GB disk | Spot | Austria | In stock | $0.02 | $0.02 | 21:07 UTCfresh | Receipt |
| GTX 1080marketplace floor | 1× GTX 108012 vCPU · 15.5 GB RAM · 649 GB disk | Spot | Austria | In stock | $0.02 | $0.02 | 21:06 UTCfresh | Receipt |
| GTX 1660 Timarketplace floor | 1× GTX 1660 Ti4 vCPU · 15.5 GB RAM · 175 GB disk | Spot | Norway | In stock | $0.02 | $0.02 | 21:07 UTCfresh | Receipt |
| GTX 1660 | 1× GTX 16604 vCPU · 7.7 GB RAM · 669.4 GB disk | Spot | Vietnam | In stock | $0.03 | $0.03 | 21:05 UTCfresh | Receipt |
| GTX 1660 Ti | 1× GTX 1660 Ti4 vCPU · 7.7 GB RAM · 679.4 GB disk | Spot | Vietnam | In stock | $0.03 | $0.03 | 21:05 UTCfresh | Receipt |
| RTX 3060marketplace floor | 1× RTX 306031.1 GB RAM · 450 GB disk | Spot | Japan | In stock | $0.03 | $0.03 | 21:05 UTCfresh | Receipt |
| GTX 1080 Timarketplace floor | 1× GTX 1080 Ti12 vCPU · 62.7 GB RAM · 50 GB disk | Spot | Japan | In stock | $0.03 | $0.03 | 21:05 UTCfresh | Receipt |
| GTX 1060 | 1× GTX 10602 vCPU · 23.4 GB RAM · 988 GB disk | Spot | France | In stock | $0.03 | $0.03 | 21:05 UTCfresh | Receipt |
| GTX 1660 | 2× GTX 166028 vCPU · 31.2 GB RAM · 800 GB disk | Spot | France | In stock | $0.03 | $0.03 | 21:06 UTCfresh | Receipt |
| GTX 1660 | 1× GTX 166014 vCPU · 15.6 GB RAM · 400 GB disk | Spot | France | In stock | $0.03 | $0.03 | 21:06 UTCfresh | Receipt |
| T4marketplace flooroutlier · suspended | 7× T456 vCPU · 94 GB RAM · 1,690 GB disk | Spot | Korea | In stock | $0.03 | $0.03 | 21:07 UTCfresh | Receipt |
| T4 | 2× T416 vCPU · 26.9 GB RAM · 482.9 GB disk | Spot | Korea | In stock | $0.03 | $0.03 | 21:07 UTCfresh | Receipt |
| T4 | 1× T48 vCPU · 13.4 GB RAM · 241.4 GB disk | Spot | Korea | In stock | $0.03 | $0.03 | 21:07 UTCfresh | Receipt |
| T4 | 4× T432 vCPU · 53.7 GB RAM · 965.7 GB disk | Spot | Korea | In stock | $0.03 | $0.03 | 21:07 UTCfresh | Receipt |
| T4 | 3× T424 vCPU · 40.3 GB RAM · 724.3 GB disk | Spot | Korea | In stock | $0.03 | $0.03 | 21:07 UTCfresh | Receipt |
| Quadro P2000marketplace floor | 4× Quadro P200020 vCPU · 27.3 GB RAM · 128.8 GB disk | On-demand | US | In stock | $0.03 | $0.03 | 21:05 UTCfresh | Receipt |
| Quadro P2000 | 2× Quadro P200010 vCPU · 13.7 GB RAM · 64.4 GB disk | On-demand | US | In stock | $0.03 | $0.03 | 21:05 UTCfresh | Receipt |
6,503 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: pricing page states 'No minimum hours, no rounding up'
- Stopped instance
- GPU charge stops; container disk keeps being billed per second at the host's storage rate until the instance is destroyed; docs note stopped-instance storage rates can be higher than running rates
- Egress
- metered: per byte, upload and download, host-set; internet cost in the public search API ranged 0 to 68.27 USD/TB across sampled offers (median 2.67 USD/TB)
- Public IPv4
- shared/NAT by default (internal ports mapped to random external ports on the host IP); some hosts list static IPs
- Billing model
- prepaid credits
- Minimum deposit
- 5 USD
- Free credits
- not published
Account
- Payment
- card · crypto (yes: via BitPay and Crypto.com (Coinbase Commerce legacy); crypto purchases are not refundable)
- Sign-up
- email, password, email verification
- Identity check
- none
- Support
- website chat, discord, contact form, email ([email protected] for compliance)
- Enterprise support
- not published
- SLA
- not published
- Status page
- status.vast.ai ↗
- Published uptime
- Status page on 2026-09-29: vast.ai 99.95%, console 99.75%, serverless 99.75% over the last 30 days
Infrastructure
- Regions
- United States, Canada, Czechia, South Korea, Taiwan, United Kingdom, Bulgaria, Japan, Romania, Germany, Spain, South Africa, Poland, China, Belgium, Sweden, Argentina, Ireland, Ukraine, Finland, Netherlands, France, Estonia, Slovakia, Norway, Saudi Arabia, Denmark, Thailand, India, Australia, Russia, Hungary, Slovenia, Italy, Croatia, Brazil, Singapore, Vietnam, Turkey, Iceland, Greece, Switzerland, Israel, Oman, Peru, Indonesia, Latvia, Austria, Hong Kong, United Arab Emirates, Mexico, Sri Lanka
- Countries
- United States, Canada, Czechia, South Korea, Taiwan, United Kingdom, Bulgaria, Japan, Romania, Germany, Spain, South Africa, Poland, China, Belgium, Sweden, Argentina, Ireland, Ukraine, Finland, Netherlands, France, Estonia, Slovakia, Norway, Saudi Arabia, Denmark, Thailand, India, Australia, Russia, Hungary, Slovenia, Italy, Croatia, Brazil, Singapore, Vietnam, Turkey, Iceland, Greece, Switzerland, Israel, Oman, Peru, Indonesia, Latvia, Austria, Hong Kong, United Arab Emirates, Mexico, Sri Lanka
- Platform
- api, cli, serverless, jupyter, ssh, custom_docker_image, persistent_volumes, spot_or_interruptible · SDK : python · Terraform : — · Kubernetes : no · interconnexion : InfiniBand on dedicated clusters (sales-led) · NVLink presence and bandwidth vary by host offer · clusters : Dedicated multi-node clusters with InfiniBand via sales form (H200 140GB SXM, H100 80GB SXM, A100 80GB SXM or other · 0-31, 32-127, 128+ GPUs)
- Storage
- ephemeral · host-set; storage_cost in the public search API ranged 0.0013 to 1.00 USD/GB-month across sampled offers (median 0.20) · Container disk: size fixed at creation (default minimum 10 GB), cannot be resized, deleted with the instance, billed while the instance existsvolume · host-set, no list price published · Local volumes tied to one physical machine; reattachable only on the same host, cannot migrate
- Migration out
- Data lives on host-local container disks or local volumes that cannot move between machines; users copy data out via Cloud Sync (S3, Google Drive, Backblaze, Dropbox), instance-to-instance copy, CLI copy or scp before destroying instances. Workloads are standard Docker images, so moving to another provider means re-pulling the same image elsewhere. Unspent card credits can be refunded on request; crypto credits cannot.
Security & compliance
- SOC 2
- Type II
- ISO 27001
- not published for Vast.ai itself; Secure Cloud datacenter partners may hold ISO 27001
- HIPAA
- Supported on Secure Cloud tier; BAAs for qualifying customers
- GDPR / DPA
- Data Processing Agreement governs data handling (linked from site footer)
- Other
- SOC 3 report on request; SOC 2 Type 2 report under NDA; About page timeline (2025) says SOC 2 Type I
- Legal entity
- Vast.ai Inc.
- Funding
- Private; incorporated as a Delaware C corporation on 28 June 2016, launched September 2018; funding not published on its site
GPUBank market study, read on the provider's own pages · 24 pages read · facts marked "not published" were looked for and not foundPages read ↓
From the study · assessmentVast.ai, in short
What stood out when we read the documents: strengths, weaknesses, who it suits, and what we could not find.
Strengths
- Very wide catalogue: 81 GPU cards on the pricing page (site claims 68+ GPU types), from GTX 1060 to B300, with 'from' and median prices visible without an account
- Low entry prices: H100 SXM from 1.73 USD/GPU-h (median 2.13), RTX 4090 from 0.14 (median 0.49), RTX 5090 from 0.33 (median 0.61) on 2026-09-29
- Interruptible bidding: docs cite savings of 50-80%; lowest H100 SXM minimum bid seen in the public API was 0.687 USD/GPU-h
- Per-second billing with no minimum and a 5 USD minimum deposit
- Crypto top-ups via BitPay and Crypto.com as well as cards
- Public, unauthenticated offer search API plus CLI and Python SDK; serverless endpoints priced at the same rate as instances
- SOC 2 Type 2 (report under NDA), SOC 3 on request, HIPAA BAAs on Secure Cloud
Weaknesses
- No fixed list price: every rate is host-set and changes hourly; the headline 'from' price is a single cheapest offer
- Storage and bandwidth are billed separately at host-set rates (bandwidth up to 68.27 USD/TB in sampled offers) and bandwidth is not shown in the $/hr figure
- Stopped instances keep paying storage; the GPU is released and may be rented by someone else
- Volumes are local to one machine and reserved instances cannot migrate between hosts
- No refund of spent credits and no refund of credits bought with crypto
- Community hosts have no formal security certification; ISO 27001 applies only to Secure Cloud partners
- No published SLA for standard instances; Linux Docker only (VMs on a subset of machines), no Windows
Ideal for
- Price-sensitive researchers and hobbyists who can compare offers and tolerate host variability
- Fault-tolerant batch jobs that can use interruptible bids
- Users who want consumer GPUs (RTX 30/40/50 series) or older cards not sold by datacenter clouds
- Teams that want to automate procurement through a public offer API
Not published
- Reserved (1/3/6 month) per-GPU prices are not published
- only 'up to 50%' and a worked example (on-demand 720 USD/month to 576 USD/month for 1 month) in docs
- Secure Cloud price levels per model are not published separately on the pricing page
- Interruptible prices are a live API snapshot (lowest min_bid among up to 64 returned offers
- the unauthenticated API returns at most 64 offers per query), not a published list
- Terraform provider and Kubernetes support: not found in the official docs index
- Number of templates not published as a figure
- Snapshot feature not documented in the pages read
- Funding and investors not published on vast.ai
- Named regions do not exist
- country list derived from host geolocation in the sampled API offers, not an official region list
- SLA for standard instances not published
- reserved docs refer to an 'implicit or explicit Service Level Agreement' per machine without figures
Observations
Pricing page lists 81 GPU cards grouped by architecture (11 groups), each with 'from' price and median per hour, updated hourly, visible without login
Two cards (GB10, GTX 1650) show only an indicative '~' price with 'No current offers'
Reserved tier routes to 'Contact Sales' on the pricing page; terms 1, 3 or 6 months, 'up to 50% off'
Home page headline 'SOC 2 Certified', claims 20,000+ GPUs, 40+ data centers, 68+ GPU types
Pricing calculator requires selecting a GPU; no static reserved or interruptible price table
Pages read
- vast.ai/pricing — 81 GPU cards with from/median prices, instance types, per-second billing, no minimum
- vast.ai/ — Scale claims, $5 start, CLI/SDK/API, serverless, clusters
- vast.ai/pricing/gpu/H100-SXM — Confirms per-GPU hourly 'from' price and hourly update
- console.vast.ai/api/v0/search/asks/ — Interruptible minimum bids, storage and bandwidth rate ranges, host countries
- docs.vast.ai/guides/instances/pricing — Cost components, reserved up to 50%, interruptible 50%+, credits upfront
- docs.vast.ai/guides/instances/choosing/instance-types — On-demand, reserved, interruptible behaviour
- docs.vast.ai/guides/reference/faq/rental-types — Interruptible bidding, 50-80% savings
- docs.vast.ai/guides/instances/choosing/reserved-instances — Terms, refunds, 5% cancellation fee, no host migration
- docs.vast.ai/guides/reference/billing — Prepaid credits, Stripe cards, BitPay and Crypto.com, refunds, storage and bandwidth billing
- docs.vast.ai/guides/reference/faq/billing — Card and email verification for card billing, no PayPal
- docs.vast.ai/guides/get-started/quickstart — Sign-up steps, email verification, 5 USD minimum deposit
- docs.vast.ai/guides/instances/storage/types — Container storage and volumes, stopped-instance storage billing
- docs.vast.ai/guides/instances/storage/volumes — Local volume limits
- docs.vast.ai/guides/reference/faq/networking — Port mapping on shared public IP
- docs.vast.ai/guides/serverless/pricing — Serverless billed at instance rates per worker state
- docs.vast.ai/guides/instances/virtual-machines — VM support on a subset of machines
- docs.vast.ai/guides/reference/faq/security — Secure Cloud definition, isolation
- docs.vast.ai/guides/reference/faq/general — Linux Docker only, SSH/Jupyter, Secure Cloud filter
- vast.ai/compliance — SOC 2 Type 2, SOC 3, HIPAA, DPA, Secure Cloud partner rules
- vast.ai/products/clusters — Sales-led InfiniBand clusters, 24/7 support claim
- vast.ai/about — Founding, HQ, offices
- vast.ai/terms — Legal entity Vast.ai Inc., refund policy, export controls
- status.vast.ai/ — 30-day uptime figures
- docs.vast.ai/llms.txt — Check for Terraform/Kubernetes docs (none found
AlternativesProviders people compare with it
RunpodOperator · 41 GPU types · H100 $2.69Trust 95
Omega GradientMarketplace · 43 GPU types · H100 $2.63not scored
EmpirioLabs AIOperator · 26 GPU types · H100 $3.80not scored
Theta EdgeCloudMarketplace · 30 GPU types · H100 $2.29not scored
RuncrateReseller · 20 GPU types · H100 $2.83not scoredVast.ai · comparisonsVast.ai against other clouds
Questions about Vast.ai
What GPUs does Vast.ai offer?
Vast.ai has 91 GPU types with a fresh offer today, for example H100 from $2.27, A100 80GB from $0.60, RTX PRO 6000 from $0.83, H200 from $2.76, L40S from $0.34, B200 from $7.50 per GPU-hour on-demand. The full list with prices and market medians is on this page.
How does Vast.ai bill?
Billing granularity: per second. Minimum charge: None: pricing page states 'No minimum hours, no rounding up' Stopped instance: GPU charge stops; container disk keeps being billed per second at the host's storage rate until the instance is destroyed; docs note stopped-instance storage rates can be higher than running rates
How trustworthy is Vast.ai?
Its Trust Score is 78/100 (v0.1), read from its own documents. Strong on company & track record, pricing & billing transparency. Weaker on infrastructure control, reliability signals. The score measures transparency and published guarantees, not customer satisfaction; every criterion and its source is listed above.
Where are Vast.ai's data centres?
52 countries: United States, Canada, Czechia, South Korea, Taiwan, United Kingdom, Bulgaria, Japan, Romania, Germany, Spain, South Africa, Poland, China, Belgium, Sweden, Argentina, Ireland, Ukraine, Finland, Netherlands, France, Estonia, Slovakia, Norway, Saudi Arabia, Denmark, Thailand, India, Australia, Russia, Hungary, Slovenia, Italy, Croatia, Brazil, Singapore, Vietnam, Turkey, Iceland, Greece, Switzerland, Israel, Oman, Peru, Indonesia, Latvia, Austria, Hong Kong, United Arab Emirates, Mexico, Sri Lanka.
How do you pay Vast.ai, and are there credits?
Payment by card, crypto. No sign-up credit is published. Identity check: none.