Jarvislabs review
Updated September 28, 2026 · DeployCue Editorial Team · Facts checked September 26, 2026
We earn a commission when you sign up through links on this site. Partner status can affect where a brand appears in our rankings. Partners with a live offer in your country come first, then every other brand by score.
Our verdict
Jarvislabs is a GPU cloud for developers who want current datacenter parts without a long commitment, with GPU hours at $0.41 and billing granularity of per minute. Buyers who need a longer card catalogue or live chat and phone support should weigh ComputePrices.com and Salad instead.
Key facts
- Best for
- Per-minute GPU rentals for H100, H200 and A100 class training and fine-tuning work.
- GPU models
- H200, H100, RTX Pro 6000, A100 80GB, A100 40GB, A30, L4↗
- GPU hour from
- $0.41↗
- Billing granularity
- Per minute↗
- Interconnect
- InfiniBand↗
- Legal entity
- Jarvis Labs Ai Private Limited↗
- Headquarters
- India↗
- Founded
- 2019↗
- Support channels
- Email, Help-centre↗
- Pricing model
- Hourly billing
- Official website
- jarvislabs.ai
Facts checked September 26, 2026, 5 sources
Pros and cons
Pros
- Entry GPU hours are $0.41, below the €0.63 for providers on this site.
- Granularity is per minute on a hourly model, so short runs stop paying when the card is released.
- The catalog spans H200, H100 and RTX Pro 6000, pairing recent training parts with lighter inference cards.
- Cluster nodes use InfiniBand, which suits multi-GPU training jobs.
Cons
- Support runs on email and help centre, with no live chat or phone channel in the mix.
- The card list is shorter than ComputePrices.com, which tracks H100 SXM, A100 SXM and H200.
- Salad lists NVIDIA RTX 3060, NVIDIA RTX 3070 and NVIDIA RTX 3070 Ti, adding consumer and selected AMD hardware that Jarvislabs does not offer.
Who it suits
- Developers training or fine-tuning on H200, H100 and A100 80GB hardware without a reserved term.
- Small teams that want $0.41 entry pricing and InfiniBand cluster interconnects.
- Users happy to work through email and help centre rather than live chat.
Who should look elsewhere
- ComputePrices.com for anyone who needs the widest hardware range, H100 SXM, A100 SXM and H200.
- SberCloud for buyers wanting B300 and A40 next to H100 and A100, as listed in B300, H100 and A100.
- Salad for jobs that can run on consumer and AMD cards, covered by NVIDIA RTX 3060, NVIDIA RTX 3070 and NVIDIA RTX 3070 Ti.
Core features
The catalog spans H200, H100 and RTX Pro 6000, so one account holds both recent training parts and lighter inference cards. Cluster nodes are linked with InfiniBand, so multi-GPU work gets direct GPU-to-GPU bandwidth rather than shared Ethernet. Single cards and clusters come from the same pricing page.
The company behind it
The provider is Jarvis Labs Ai Private Limited, headquartered in India and founded in 2019. Support runs through email and help centre, so questions go by email or through the help centre rather than a live chat or a phone line, which matters during a long training run.
How Jarvislabs compares
Jarvislabs vs RunPod9.1 vs 9.5
Jarvislabs vs ScalaHosting9.1 vs 8.3
Jarvislabs vs Contabo9.1 vs 9.2
Jarvislabs vs Kamatera9.1 vs 9.5
Jarvislabs vs Cloudways9.1 vs 9.2
Jarvislabs vs PlanetHoster9.1 vs 8.7
Jarvislabs vs Acronis9.1 vs 9.9
Jarvislabs vs Aruba9.1 vs 9.3
Jarvislabs vs Decodo9.1 vs 9.1
Jarvislabs vs Hostinger9.1 vs 9.1
Jarvislabs vs MarsProxies9.1 vs 9.0
Jarvislabs vs hide.me9.1 vs 8.7
Jarvislabs vs Proton Drive9.1 vs 8.3
Where Jarvislabs is available
- Available
- Argentina, Australia, Austria, Belgium, Bolivia, Brazil, Bulgaria, Canada, Chile, Colombia, Costa Rica, Cyprus, Czechia, Denmark, Estonia, Finland, France, Germany, Ghana, Greece, Guernsey, Hungary, India, Indonesia, Ireland, Isle of Man, Israel, Italy, Japan, Jersey, Kenya, Latvia, Lithuania, Luxembourg, Mexico, Nepal, Netherlands, New Zealand, Nigeria, Norway, Poland, Portugal, Puerto Rico, Romania, Serbia, Singapore, Slovakia, Slovenia, South Africa, South Korea, Spain, Sri Lanka, Sweden, Taiwan, Thailand, Tunisia, United Arab Emirates, United Kingdom, United States, Virgin Islands, British, Zambia
Countries from the brand (Source, checked September 26, 2026)
Frequently asked questions
What does a GPU hour cost at entry?+
GPU hours come in at $0.41, below the €0.63 across tracked providers. On that measure Jarvislabs ranks 22 of 95 brands compared. Larger cards and multi-GPU clusters cost more than the entry rate.
How fine is the billing meter?+
The pricing model is hourly and granularity is per minute, so short runs stop paying once the card is released. Providers that meter per second sit at 26.4% of tracked sites, so a finer meter exists elsewhere.
Which GPUs are on offer?+
The catalog is H200, H100 and RTX Pro 6000, covering recent training parts and lighter inference cards in one account. Compare the lists: ComputePrices.com tracks far more parts, while Salad runs consumer cards and selected AMD classes for jobs that avoid datacenter silicon.
What support do I get?+
Support runs on email and help centre. Email and a help centre answer most questions, with no live chat or phone channel in the mix, so teams expecting real-time help during a long training run should compare providers with wider support.
Who runs Jarvislabs and where?+
The provider is Jarvis Labs Ai Private Limited, headquartered in India and founded in 2019. Contracts and invoices sit with that entity, which matters for buyers who prefer a counterparty in their own market.
Are monthly or reserved plans sold?+
Capacity is metered as hourly usage rather than a subscription. Across tracked sites, 90.8% offer monthly plans and 15.6% sell reserved terms, so Jarvislabs suits buyers who want to stop without a commitment.
Can I run multi-GPU training?+
Cluster nodes are linked with InfiniBand, giving multi-GPU training direct GPU-to-GPU bandwidth instead of shared Ethernet. Single cards come from the same catalog, H200, H100 and RTX Pro 6000, so small tests and full clusters draw on one pricing page.
Sources
- docs.jarvislabs.ai checked September 26, 2026
Backs: Pricing model
“pay-as-you-go flexibility”
- jarvislabs.ai/contactus checked September 26, 2026
Backs: Headquarters, Support channels
“New Delhi 110044, India”
- jarvislabs.ai/pricing checked September 26, 2026
Backs: Billing granularity, GPU hour from, GPU models
“On-demand instances are billed per minute”
- jarvislabs.ai/products/clusters checked September 26, 2026
Backs: Interconnect
“NVIDIA H200 nodes with eight GPUs and InfiniBand”
- cbinsights.com/company/jarvislabs checked September 26, 2026
Backs: Founded
“Jarvislabs was founded in 2019”
- thecompanycheck.com/company/jarvis-labs-ai-private-limite... checked September 26, 2026
Backs: Legal entity
“Jarvis Labs Ai Private Limited”
Compiled from 6 source pages, checked September 26, 2026. We did not open an account.
Not confirmed yet: Egress fee.
Related by the facts
Jarvislabs
Per-minute GPU rentals for H100, H200 and A100 class training and fine-tuning work.