GPU Rental Pricing
Price per GPU-hour for every GPU model on the Lium marketplace. A range is the spread of live asks across the pods rentable right now; a single figure is the reference price of a model with no pod available at the moment. Every pod is billed per second with no minimum.
Data center
| GPU | $ / GPU / hour | Available now |
|---|---|---|
| B300 SXM6 | $7.99 – $12.95 | 5 pods · 19 GPUs |
| GB300 | $8.00 reference | — |
| B200 | $5.50 – $7.52 | 7 pods · 14 GPUs |
| H200 | $3.73 – $5.76 | 5 pods · 12 GPUs |
| H200 NVL | $2.90 reference | — |
| H100 80GB HBM3 | $1.39 reference | — |
| H100 PCIe | $1.30 reference | — |
| H100 NVL | $1.11 reference | — |
| H800 80GB HBM3 | $0.88 reference | — |
| H800 NVL | $0.80 reference | — |
| H800 PCIe | $0.80 reference | — |
| A100-SXM4-80GB | $1.00 – $1.22 | 2 pods · 9 GPUs |
| A100 80GB PCIe | $0.45 reference | — |
| L40S | $0.38 | 3 pods · 9 GPUs |
| L40 | $0.33 reference | — |
| A800 80GB PCIe | $0.31 reference | — |
| CMP 170HX | $0.30 reference | — |
| A10 | $0.20 reference | — |
| Tesla V100 | $0.17 reference | — |
| T4 | $0.15 reference | — |
| A40 | $0.12 reference | — |
| L4 | $0.11 reference | — |
| A30 | $0.10 reference | — |
| Tesla P100 | $0.08 reference | — |
| Tesla M40 | $0.08 reference | — |
| Tesla P40 | $0.07 reference | — |
Professional
Consumer
| GPU | $ / GPU / hour | Available now |
|---|---|---|
| RTX 2080 SUPER | $0.74 reference | — |
| RTX 5090 | $0.50 – $0.85 | 8 pods · 37 GPUs |
| GTX 1070 | $0.37 reference | — |
| RTX 4090 | $0.45 – $0.58 | 8 pods · 14 GPUs |
| RTX 4070 Ti SUPER | $0.18 reference | — |
| RTX 4080 SUPER | $0.16 reference | — |
| RTX 4080 | $0.16 reference | — |
| RTX 3090 Ti | $0.16 reference | — |
| RTX 3090 | $0.22 | 2 pods · 7 GPUs |
| RTX 5080 | $0.15 reference | — |
| RTX 5070 Ti | $0.12 reference | — |
| RTX 5060 Ti | $0.12 reference | — |
| RTX 5060 | $0.11 reference | — |
| RTX 4090 D | $0.11 reference | — |
| RTX 4070 Ti | $0.11 reference | — |
| RTX 4060 Ti | $0.11 reference | — |
| RTX 3080 Ti | $0.10 reference | — |
| RTX 5070 | $0.09 reference | — |
| RTX 4070 SUPER | $0.09 reference | — |
| RTX 3080 | $0.09 reference | — |
| RTX 3070 Ti | $0.09 reference | — |
| GTX 1060 | $0.09 reference | — |
| RTX 4070 | $0.08 reference | — |
| RTX 3060 Laptop GPU | $0.08 reference | — |
| RTX 2080 Ti | $0.08 reference | — |
| RTX 3070 | $0.07 reference | — |
| RTX 2070 SUPER | $0.07 reference | — |
| GTX 1660 Ti | $0.07 reference | — |
| GTX 1660 | $0.07 reference | — |
| RTX 4060 | $0.06 reference | — |
| RTX 3060 Ti | $0.06 reference | — |
| RTX 3060 | $0.06 reference | — |
| RTX 2060 SUPER | $0.06 reference | — |
| RTX 3050 | $0.05 reference | — |
| GTX 1660 SUPER | $0.05 reference | — |
| RTX 2060 | $0.04 reference | — |
| GTX 1080 Ti | $0.04 reference | — |
| GTX 1070 Ti | $0.04 reference | — |
| GTX 1080 | $0.03 reference | — |
Frequently asked questions
What do the prices on this page mean?
Each figure is the price of one GPU for one hour, set by the provider that owns the machine. A range is the spread of live asks across the pods rentable right now. A single figure is one live ask or, where the availability column shows a dash, the reference price of a model with no pod free. 11 of 92 models have a pod free right now, 153 GPUs in total.
How is a Lium pod billed?
Per second, with no minimum and no reservation. You need a balance that covers 15 minutes of the pod to start it; billing stops when you delete the pod or its scheduled end time arrives.
Where can a script or an aggregator read these prices?
The same table is served as JSON at lium.io/pricing.json, every model page is served as markdown to a client that asks for text/markdown, and the live node list is at lium.io/api/public/v1/nodes. None of them needs an account.
Do the prices include storage or a platform fee?
The price shown is the whole hourly price of the pod, including its local disk; the provider receives 95 % of what you pay for the rental. A persistent volume, if you attach one, is billed separately: storage per GB-hour plus data read out of the volume, both at cost.
For AI agents — CLI install, agent skill, headless signup: llms-full.txt