Cheapest on-demand GPU rental · October 23, 2026
Cheapest on-demand GPU rental, by provider type
The cheapest card for the memory you need, on-demand rate cards only: community capacity, shared GPUs and marketplace floors are shown apart. Listed rates, not availability. The same, in words ↓
Cheapest on-demand rate per GPU-hour
Cheapest card for the memory you need
As of , the cheapest on-demand rate per GPU-hour among cards with at least 24 GB: $0.16 for the NVIDIA V100 32GB (32 GB) on Geodd; at least 48 GB: $0.35 for the NVIDIA RTX A6000 (48 GB) on Thunder Compute; at least 80 GB: $0.65 for the NVIDIA RTX PRO 6000 Blackwell (96 GB) on Hostinger; at least 141 GB: $1.71 for the AMD Instinct MI300X (192 GB) on TensorWave. The largest step up is from 96 GB to 141 GB of memory: the cheapest rate rises from $0.65 (NVIDIA RTX PRO 6000 Blackwell) to $1.71 (AMD Instinct MI300X), 2.6 times as much.
| Memory per GPU | Card | Cheapest on-demand | Provider | Captured |
|---|---|---|---|---|
| At least 24 GB | NVIDIA V100 32GB (32 GB) | $0.16 |  Geodd | October 23, 2026, 21:40 UTC |
| At least 48 GB | NVIDIA RTX A6000 (48 GB) | $0.35 |  Thunder Compute | October 23, 2026, 21:40 UTC |
| At least 80 GB | NVIDIA RTX PRO 6000 Blackwell (96 GB) | $0.65 from · |  Hostinger | October 23, 2026, 21:40 UTC |
| At least 141 GB | AMD Instinct MI300X (192 GB) | $1.71 from · |  TensorWave | October 23, 2026, 21:40 UTC |
Which card for which job
The table does not tell you which GPU to rent; the card decides that, not the provider. As a rule of thumb from the numbers above: training and large fine-tunes need the memory of an H100, H200 or B200; smaller fine-tunes and most inference of 7–70B models fit an A100 80GB, L40S or RTX 6000 Ada; small-model inference, image generation and hobby work run on an L4, RTX 4090 or RTX 5090. Within a card, the cheapest listed rate is usually a marketplace floor or a community tier; the hyperscaler rate is what you pay for the same card with enterprise contracts and regions around it.
The same, in words
As of , the cheapest listed on-demand rates are:
- NVIDIA GB300 NVL72: $18.00 on Oracle Cloud (OCI) among hyperscalers; $12.29 on Verda among specialist GPU clouds.
- NVIDIA GB200 NVL72: $16.00 on Oracle Cloud (OCI) among hyperscalers; $6.94 on Bitdeer AI (from) among specialist GPU clouds.
- NVIDIA B300: $15.00 on Oracle Cloud (OCI) among hyperscalers; $5.89 on Deep Infra among specialist GPU clouds; $6.94 on RunPod (Community Cloud) as community capacity or a marketplace floor.
- AMD Instinct MI350X: $4.00 on CloudRift among specialist GPU clouds.
- NVIDIA B200: $14.00 on Oracle Cloud (OCI) among hyperscalers; $4.11 on Beam (from) among specialist GPU clouds; $3.75 on packet.ai (Dynamic) as a shared GPU, a card other tenants use too.
- NVIDIA H200 NVL: $3.36 on IONOS Cloud among specialist GPU clouds; $3.92 on Vast.ai (Multi-GPU machines, floor) as community capacity or a marketplace floor.
- AMD Instinct MI355X: $8.60 on Oracle Cloud (OCI) among hyperscalers; $2.95 on TensorWave (from) among specialist GPU clouds.
- NVIDIA GH200: $2.29 on Lambda among specialist GPU clouds.
- AMD Instinct MI325X: $2.25 on Bentaus among specialist GPU clouds.
- NVIDIA H200: $7.91 on Amazon Web Services (us-east-1) among hyperscalers; $2.09 on Beam (from) among specialist GPU clouds; $3.59 on GPU.ai (from) as community capacity or a marketplace floor.
- NVIDIA H100 NVL: $6.98 on Microsoft Azure (eastus) among hyperscalers; $2.06 on GPU.ai (from) among specialist GPU clouds; $2.59 on RunPod (Community Cloud) as community capacity or a marketplace floor.
- NVIDIA H100 PCIe: $1.83 on Beam (from) among specialist GPU clouds; $1.99 on RunPod (Community Cloud) as community capacity or a marketplace floor.
- NVIDIA H100 SXM: $6.88 on Amazon Web Services (us-east-1) among hyperscalers; $1.79 on Cudo Compute (no-kristiansand-1, GPU only) among specialist GPU clouds; $2.69 on RunPod (Community Cloud) as community capacity or a marketplace floor.
- NVIDIA H100 80GB (form factor not stated): $1.71 on Gcore (Everywhere Inference) among specialist GPU clouds.
- AMD Instinct MI300X: $6.00 on Microsoft Azure (eastus2) among hyperscalers; $1.71 on TensorWave (from) among specialist GPU clouds.
- NVIDIA RTX PRO 6000 Blackwell Max-Q: $1.80 on EmpirioLabs among specialist GPU clouds; $1.47 on Vast.ai (Multi-GPU machines, floor) as community capacity or a marketplace floor.
- NVIDIA A10G: $1.01 on Amazon Web Services (us-east-1) among hyperscalers; $1.00 on Hugging Face (Spaces hardware) among specialist GPU clouds.
- NVIDIA A100 80GB: $3.43 on Amazon Web Services (us-east-1) among hyperscalers; $1.05 on CloudRift among specialist GPU clouds; $0.89 on packet.ai (Dynamic) as a shared GPU, a card other tenants use too.
- NVIDIA RTX PRO 5000 Blackwell: $0.96 on RunPod (Secure Cloud) among specialist GPU clouds; $0.74 on Vast.ai (floor) as community capacity or a marketplace floor.
- NVIDIA L40: $0.78 on GPU.ai (from) among specialist GPU clouds; $0.69 on GPU.ai (from) as community capacity or a marketplace floor.
- NVIDIA RTX 6000 Ada: $0.65 on GPU.ai (from) among specialist GPU clouds; $0.74 on GPU.ai (from) as community capacity or a marketplace floor.
- NVIDIA RTX PRO 6000 Blackwell: $3.36 on Amazon Web Services (us-east-1) among hyperscalers; $0.65 on Hostinger (from) among specialist GPU clouds; $0.66 on packet.ai (Dynamic) as a shared GPU, a card other tenants use too.
- NVIDIA A100 40GB: $2.74 on Amazon Web Services (us-east-1) among hyperscalers; $0.68 on Trooper.AI (from) among specialist GPU clouds; $0.63 on Vast.ai (Multi-GPU machines, floor) as community capacity or a marketplace floor.
- NVIDIA L40S: $1.86 on Amazon Web Services (us-east-1) among hyperscalers; $0.63 on CloudRift among specialist GPU clouds; $0.79 on GPU.ai (from) as community capacity or a marketplace floor.
- NVIDIA RTX 5090: $0.55 on HexGrid (Secure Cloud) among specialist GPU clouds; $0.50 on Salad (High priority) as community capacity or a marketplace floor.
- NVIDIA RTX 5000 Ada Generation: $0.83 on RunPod (Secure Cloud) among specialist GPU clouds; $0.49 on RunPod (Community Cloud) as community capacity or a marketplace floor.
- NVIDIA A40: $0.59 on GPU.ai (from) among specialist GPU clouds; $0.35 on RunPod (Community Cloud) as community capacity or a marketplace floor.
- NVIDIA A10: $2.00 on Oracle Cloud (OCI) among hyperscalers; $0.34 on Intelion Cloud (GPU only) among specialist GPU clouds.
- NVIDIA RTX PRO 4500 Blackwell: $2.52 on Amazon Web Services (us-east-1) among hyperscalers; $0.72 on GPU.ai (from) among specialist GPU clouds; $0.34 on RunPod (Community Cloud) as community capacity or a marketplace floor.
- NVIDIA RTX 4090: $0.39 on CloudRift among specialist GPU clouds; $0.33 on Salad (High priority) as community capacity or a marketplace floor.
- NVIDIA RTX A6000: $0.35 on Thunder Compute among specialist GPU clouds; $0.33 on RunPod (Community Cloud) as community capacity or a marketplace floor.
- NVIDIA A30: $0.32 on SynpixCloud among specialist GPU clouds.
- NVIDIA GeForce RTX 4080 SUPER: $0.28 on RunPod (Community Cloud) as community capacity or a marketplace floor.
- NVIDIA L4: $0.56 on Google Cloud (us-east4, GPU only) among hyperscalers; $0.27 on Hikube (GPU only) among specialist GPU clouds; $0.36 on Vast.ai (floor) as community capacity or a marketplace floor.
- NVIDIA Quadro RTX 6000: $0.27 on centron (GPU only) among specialist GPU clouds.
- NVIDIA GeForce RTX 5080: $0.36 on SwissGPU among specialist GPU clouds; $0.26 on Salad (High priority) as community capacity or a marketplace floor.
- NVIDIA T4: $0.35 on Google Cloud (us-east5, GPU only) among hyperscalers; $0.24 on immers.cloud among specialist GPU clouds.
- NVIDIA RTX 2000 Ada: $0.24 on GPU.ai (from) among specialist GPU clouds.
- NVIDIA GeForce RTX 4080: $0.30 on SwissGPU among specialist GPU clouds; $0.23 on Salad (High priority) as community capacity or a marketplace floor.
- NVIDIA V100: $2.48 on Google Cloud (us-west4, GPU only) among hyperscalers; $0.23 on Verda among specialist GPU clouds; $0.23 on RunPod (Community Cloud) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 5070 Ti: $0.23 on Salad (High priority) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 4070 Ti SUPER: $0.29 on Trooper.AI (from) among specialist GPU clouds; $0.21 on Salad (High priority) as community capacity or a marketplace floor.
- NVIDIA RTX 4000 Ada: $0.28 on GPU.ai (from) among specialist GPU clouds; $0.20 on RunPod (Community Cloud) as community capacity or a marketplace floor.
- NVIDIA A2: $0.20 on Infomaniak among specialist GPU clouds.
- NVIDIA GeForce RTX 3090 Ti: $0.46 on RunPod (Secure Cloud) among specialist GPU clouds; $0.19 on Salad (High priority) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 4070 Ti: $0.19 on RunPod (Community Cloud) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 4070 SUPER: $0.18 on Vast.ai (floor) as community capacity or a marketplace floor.
- NVIDIA RTX 4000 SFF Ada: $0.44 on RunPod (Secure Cloud) among specialist GPU clouds; $0.18 on RunPod (Community Cloud) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 3090: $0.30 on SwissGPU among specialist GPU clouds; $0.17 on Clore.ai (floor) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 5070: $0.17 on Salad (High priority) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 3070 Ti: $0.16 on Clore.ai (Multi-GPU machines, floor) as community capacity or a marketplace floor.
- NVIDIA RTX A5000: $0.27 on GPU.ai (from) among specialist GPU clouds; $0.16 on GPU.ai (from) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 3080 Ti: $0.15 on Salad (High priority) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 5060 Ti: $0.15 on Salad (High priority) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 4070: $0.14 on Clore.ai (Multi-GPU machines, floor) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 3080: $0.23 on SynpixCloud among specialist GPU clouds; $0.14 on Salad (High priority) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 4060 Ti: $0.14 on Salad (High priority) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 2080 Ti: $0.23 on SynpixCloud among specialist GPU clouds; $0.12 on Vast.ai (floor) as community capacity or a marketplace floor.
- NVIDIA RTX A2000: $0.12 on RunPod (Community Cloud) as community capacity or a marketplace floor.
- NVIDIA V100 32GB: $0.16 on Geodd among specialist GPU clouds; $0.12 on Clore.ai (Multi-GPU machines, floor) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 3060 Ti: $0.12 on Clore.ai (Multi-GPU machines, floor) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 3070: $0.11 on Clore.ai (Multi-GPU machines, floor) as community capacity or a marketplace floor.
- NVIDIA RTX A4000: $0.15 on centron (GPU only) among specialist GPU clouds; $0.11 on Vast.ai (floor) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 5060: $0.11 on Salad (High priority) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 2060 12GB: $0.11 on Vast.ai (Multi-GPU machines, floor) as community capacity or a marketplace floor.
- NVIDIA GeForce RTX 3060: $0.07 on Vast.ai (floor) as community capacity or a marketplace floor.
- NVIDIA GeForce GTX 1080 Ti: $0.23 on SynpixCloud among specialist GPU clouds; $0.07 on Vast.ai (floor) as community capacity or a marketplace floor.
- NVIDIA Quadro P4000: $0.07 on Vast.ai (floor) as community capacity or a marketplace floor.
- NVIDIA TITAN Xp: $0.07 on Vast.ai (Multi-GPU machines, floor) as community capacity or a marketplace floor.
- NVIDIA GeForce GTX 1660 SUPER: $0.06 on Vast.ai (floor) as community capacity or a marketplace floor.
- NVIDIA GeForce GTX 1080: $0.05 on Vast.ai (Multi-GPU machines, floor) as community capacity or a marketplace floor.
- NVIDIA GeForce GTX 1070 Ti: $0.05 on Vast.ai (Multi-GPU machines, floor) as community capacity or a marketplace floor.
We read published price lists, not availability: a listed rate can be sold out, waitlisted or absent from the region you need, and whether a card is free to rent is not something we measure.
For every GPU we track, the lowest publicly listed on-demand rate from each kind of provider: the 4 hyperscalers (AWS, Azure, Google Cloud, Oracle) and 88 specialist GPU clouds, and beside them the cheapest community capacity or marketplace floor (RunPod Community Cloud on third-party hosts, the Vast.ai floor), which can be reclaimed or moves with supply and so never counts as on-demand. Prices only — no ranking by opinion. A tag names the tier or scope; “GPU only” means CPU and memory are billed separately, “floor” means a marketplace offer that moves with supply.
On-demand list prices only: spot, reserved and negotiated rates are not compared here. Spot rates are on each GPU page (every card); the weekly H100 series is the Price Index.
Data updated: