Best GPU cloud for LLM hosting (2026), priced by the hour
thecontinuitylayer.com/tools/gpu-hostingFor a first experiment, rent a 24GB card by the hour on Runpod or Vast.ai for under a dollar. For a model you will serve all day, Lambda, Nebius or DigitalOcean give you a plain H100 at a published hourly rate, and Modal is the pick when traffic comes in bursts.
Picking the best GPU cloud for LLM work comes down to the graphics card you need, the price per hour, and how you are billed. These are the clouds that rent GPUs by the hour or the second for running open models. Prices are the ones each company publishes.
10 entries · Last reviewed 31 October 2026 · US dollars · No affiliate links
Compared at a glance
Prices and fees are the ones each vendor published on the day in the table; check the vendor before buying. Price per hour read from each vendor's own page on 2 November 2026; 7 of 10 carried a figure that day. Where a vendor showed none, the cell says “not published”.
| Tool | Best for | Price per hour | Verdict | Link |
|---|---|---|---|---|
| RunpodGPU cloud | Cheap GPU by the hour, serverless | $0.27–$7.99/hr per GPU | The default place to try an open model for a few dollars, with serverless endpoints when you outgrow a pod. | runpod.io ↗ |
| Vast.aiMarketplace | Lowest marketplace prices | not published | Usually the cheapest hour for a 24GB card, if you accept picking a host and the odd interruption. | vast.ai ↗ |
| LambdaGPU cloud | Plain NVIDIA machines for teams | $0.79–$6.69/hr per GPU | Simple, well-run data-center GPUs for teams; too much machine for hobby budgets. | lambda.ai ↗ |
| CoreWeaveGPU cloud | Whole servers and clusters | $49.24–$68.80/hr per 8-GPU server | Built for companies renting whole servers and clusters, not for a single-GPU experiment. | coreweave.com ↗ |
| Together AIGPU cloud | Managed endpoint for an open model | $3.99–$8.19/hr per GPU (clusters) | The easiest step from a per-token API to your own dedicated capacity, at a premium over raw GPU hours. | together.ai ↗ |
| Novita AIGPU cloud | API plus rentable GPUs, one account | not published | Handy if you want an open-model API and spot GPUs in one place; published rates change often. | novita.ai ↗ |
| GPU DropletsGPU cloud | GPU next to an existing app | $0.76–$4.47/hr per GPU | The sensible pick if you already run on DigitalOcean; not the cheapest H100 hour. | digitalocean.com ↗ |
| HyperbolicMarketplace | Short H100 and H200 rentals | not published | On-demand data-center GPUs without a contract, aimed at developers rather than consumer-card budgets. | hyperbolic.ai ↗ |
| ModalServerless GPU | Serve a model as code, pay while it runs | $0.000164–$0.001736 per GPU-second (about $0.59–$6.25/hr, converted) | Best for spiky or low-traffic endpoints written in Python; a rented GPU is cheaper for steady load. | modal.com ↗ |
| Nebius AI CloudGPU cloud | Recent NVIDIA GPUs, self-service | $1.55–$8.50/hr per GPU | A full AI cloud with self-service H200 and B200 access, better for sustained work than a quick test. | nebius.com ↗ |
Every gpu hosting entry
GPU cloud · Runpod
Runpod
- Price per hour
- RTX 4090 from $0.34/hr; H100 PCIe from $1.99/hr (Community Cloud)
- GPUs
- Consumer and data-center NVIDIA GPUs, from RTX 4090 to H100 and newer Blackwell cards
Builders who want a cheap GPU by the hour or a serverless endpoint for an open model
Marketplace · Vast.ai
Vast.ai
- Price per hour
- Varies; marketplace prices for each GPU change with supply and demand
- GPUs
- Marketplace of consumer and data-center GPUs from independent hosts, from older RTX cards to H100
Cost-conscious tinkerers and researchers who are comfortable picking a host themselves
GPU cloud · Lambda
Lambda
- Price per hour
- H100 SXM from $3.99/hr; B200 from $6.69/hr (per GPU, on demand)
- GPUs
- NVIDIA data-center GPUs including B200, H100, A100 and GH200, in 1 to 8 GPU instances and clusters
Teams that want plain, well-run NVIDIA machines for training or serving larger open models
GPU cloud · CoreWeave
CoreWeave
- Price per hour
- HGX H100 (8 GPUs) $49.24/hr; HGX B200 (8 GPUs) $68.80/hr on demand
- GPUs
- Large NVIDIA clusters: HGX H100, H200, B200 and GB200 NVL72 systems
Companies running open models at large scale that need whole GPU servers and clusters
GPU cloud · Together AI
Together AI
- Price per hour
- GPU clusters: H100 $3.99/hr, B200 $8.19/hr per GPU on demand
- GPUs
- NVIDIA H100, H200, B200 and GB200 as dedicated endpoints or self-service clusters
Teams that want a dedicated, managed endpoint for an open model without running servers
GPU cloud · Novita AI
Novita AI
- Price per hour
- RTX 4090 $0.67/hr on demand; spot instances about half price
- GPUs
- Consumer and data-center GPUs such as RTX 4090, RTX 5090, H100 and H200, plus a per-token model API
Developers who want both an open-model API and rentable GPUs from one account
GPU cloud · DigitalOcean
GPU Droplets
- Price per hour
- RTX 4000 Ada $0.76/hr; L40S $1.57/hr; H100 $4.41/hr per GPU on demand
- GPUs
- NVIDIA RTX 4000 Ada, L40S, H100, H200 and AMD MI300X
Developers already on DigitalOcean who want a GPU server next to their app
Marketplace · Hyperbolic
Hyperbolic
- Price per hour
- H100 SXM from $3.19/hr; H200 from $3.99/hr; B200 from $5.99/hr
- GPUs
- Rentable H100, H200 and B200 GPUs from a supplier marketplace, plus an open-model API
Developers who want on-demand data-center GPUs without a contract
Serverless GPU · Modal Labs
Modal
- Price per hour
- L4 $0.000222/sec (about $0.80/hr); A100 80GB $0.000694/sec (about $2.50/hr); H100 $0.001097/sec (about $3.95/hr)
- GPUs
- Serverless NVIDIA GPUs: T4, L4, A10, L40S, A100, H100, H200, B200 and B300
Python developers who want to serve an open model as code, paying only while it runs
GPU cloud · Nebius
Nebius AI Cloud
- Price per hour
- H100 $4.50/hr; H200 $5.40/hr; B200 $8.50/hr per GPU on demand
- GPUs
- NVIDIA H100, H200, B200 and B300 with NVLink, from single GPUs to large clusters
Teams that want a full-featured AI cloud with self-service access to recent NVIDIA GPUs
Questions people ask
- What does it cost to rent a GPU for an open model?
- A 24GB consumer card such as an RTX 4090 rents for well under a dollar an hour on the marketplaces and community tiers. An 80GB H100 runs from about $2 to $4.50 an hour on demand, depending on the cloud. Billing is usually per second or per minute, so a short test costs cents.
- Which GPU do I need for a 70B model?
- A 70B model at 4-bit needs about 40GB of GPU memory, so one 80GB H100 or H200, or two 24GB cards, is the usual choice. For 8-bit or long context, plan on 80GB or more. Smaller 27–32B models fit on a single 24GB or 48GB card.
- Serverless or a rented GPU: which is cheaper?
- Serverless billing such as Modal's or Runpod's charges only while a request runs, so it wins when traffic is light or spiky. A GPU rented by the hour is cheaper once it is busy most of the day. Marketplaces such as Vast.ai are cheapest of all, with the trade-off that a host can go offline.
How we pick: a tool is listed because people use it to run open models, never because it pays. No link on this page pays us today; if that changes it will be marked here. How links work on this site. New to open models? Start with Run your own AI.