GPU Cloud Price Report: October 2026
Month to date. 50 GPU models and 957 live offers, priced per GPU per hour, from $0.088/hr for RTX 3070. Live per-model detail is on the Aquanode GPU Index.
Data through: 2026-10-06 18:46:59 UTC. Refreshes hourly.
| GPU | Memory | Lowest per hour | Data-center lowest per hour | Median per hour | P90 per hour | Offers | Providers |
|---|---|---|---|---|---|---|---|
| RTX 3070 | 8 GB | $0.088 | n/a | $0.143 | $0.143 | 9 | 2 |
| V100 | 16 GB | $0.088 | n/a | $0.209 | $0.253 | 26 | 3 |
| RTX 3060 | 12 GB | $0.110 | n/a | $0.110 | $0.110 | 1 | 1 |
| RTX 4070 Super | 12 GB | $0.121 | n/a | $0.121 | $0.121 | 5 | 1 |
| RTX A2000 | 6 GB | $0.132 | n/a | $0.132 | $0.132 | 4 | 1 |
| RTX 3060 TI | 8 GB | $0.162 | n/a | $0.215 | $0.258 | 2 | 2 |
| RTX A5000 | 24 GB | $0.176 | $0.297 | $0.237 | $0.297 | 16 | 1 |
| RTX 3080 | 10 GB | $0.187 | n/a | $0.198 | $0.290 | 9 | 2 |
| RTX A4000 | 16 GB | $0.187 | $0.275 | $0.231 | $0.275 | 16 | 1 |
| RTX 5060 Ti | 16 GB | $0.188 | n/a | $0.273 | $0.406 | 6 | 1 |
| RTX 3090 | 24 GB | $0.198 | $0.550 | $0.297 | $0.550 | 30 | 3 |
| RTX 4000 SFF Ada | 20 GB | $0.198 | n/a | $0.198 | $0.198 | 7 | 1 |
| RTX 4070 Ti | 12 GB | $0.209 | n/a | $0.209 | $0.209 | 5 | 1 |
| RTX A4500 | 20 GB | $0.209 | $0.275 | $0.242 | $0.275 | 8 | 1 |
| RTX 4000 Ada | 20 GB | $0.220 | $0.308 | $0.287 | $0.308 | 19 | 3 |
| RTX 6000 | 24 GB | $0.254 | n/a | $0.254 | $0.254 | 1 | 1 |
| RTX 5070 | 12 GB | $0.262 | n/a | $0.334 | $0.379 | 4 | 1 |
| RTX 2000 Ada | 16 GB | $0.264 | $0.264 | $0.264 | $0.264 | 8 | 1 |
| RTX 4070 Ti Super | 16 GB | $0.264 | n/a | $0.264 | $0.275 | 43 | 1 |
| RTX 4080 | 16 GB | $0.297 | n/a | $0.405 | $0.516 | 5 | 2 |
| RTX 4080 Super | 32 GB | $0.338 | n/a | $0.442 | $0.442 | 3 | 1 |
| RTX A6000 | 48 GB | $0.363 | $0.550 | $0.594 | $0.627 | 32 | 4 |
| RTX 4090 | 24 GB | $0.374 | $0.761 | $0.614 | $0.884 | 104 | 3 |
| A30 | 24 GB | $0.385 | $0.385 | $0.385 | $0.385 | 1 | 1 |
| RTX 5080 | 16 GB | $0.411 | n/a | $0.429 | $0.641 | 14 | 2 |
| Quadro P2000 | 5 GB | $0.416 | n/a | $0.416 | $0.416 | 1 | 1 |
| RTX PRO 4000 | 24 GB | $0.441 | $0.627 | $0.627 | $0.722 | 16 | 2 |
| RTX 5070 Ti | 16 GB | $0.442 | n/a | $0.516 | $0.609 | 5 | 1 |
| L4 | 24 GB | $0.478 | $0.539 | $0.539 | $0.539 | 11 | 2 |
| A40 | 48 GB | $0.539 | $0.539 | $0.539 | $0.539 | 11 | 1 |
| RTX 5000 Ada | 32 GB | $0.539 | n/a | $0.539 | $0.539 | 4 | 1 |
| RTX 5090 | 32 GB | $0.565 | $0.904 | $0.905 | $1.35 | 102 | 3 |
| L40 | 48 GB | $0.759 | $0.850 | $0.902 | $1.11 | 30 | 3 |
| RTX PRO 4500 | 32 GB | $0.792 | $0.792 | $0.792 | $0.792 | 9 | 1 |
| RTX PRO 4500 SE | 32 GB | $0.792 | $0.792 | $0.792 | $0.792 | 8 | 1 |
| RTX 6000 Ada | 48 GB | $0.814 | $0.869 | $0.897 | $0.924 | 24 | 3 |
| RTX PRO 5000 | 48 GB | $0.862 | n/a | $0.902 | $1.17 | 18 | 2 |
| L40S | 48 GB | $0.869 | $1.07 | $1.20 | $2.91 | 39 | 5 |
| RTX PRO 4500 Blackwell | 32 GB | $1.01 | $1.01 | $1.01 | $1.01 | 4 | 1 |
| A100 | 80 GB | $1.10 | $1.49 | $1.53 | $1.75 | 46 | 5 |
| DGX A100 | 80 GB | $1.52 | $1.52 | $1.52 | $1.52 | 1 | 1 |
| RTX PRO 6000 | 96 GB | $1.54 | $1.98 | $2.05 | $2.41 | 69 | 6 |
| RTX PRO 6000 WS | 96 GB | $1.69 | n/a | $1.98 | $2.50 | 11 | 1 |
| RTX PRO 6000 SE | 96 GB | $2.15 | n/a | $2.15 | $2.15 | 1 | 1 |
| H100 | 80 GB | $2.19 | $2.78 | $3.51 | $3.84 | 68 | 8 |
| AMD MI300X | 192 GB | $2.63 | $2.63 | $2.63 | $2.76 | 9 | 2 |
| H200 | 141 GB | $3.95 | $3.98 | $5.05 | $5.94 | 60 | 5 |
| MI 350 X | 294 GB | $6.04 | $6.04 | $6.04 | $6.04 | 8 | 1 |
| B200 | 180 GB | $7.47 | $7.47 | $7.47 | $11.18 | 14 | 3 |
| B300 | 288 GB | $8.68 | $8.68 | $8.68 | $8.68 | 10 | 1 |
Trends draw on 1 daily snapshot, from 2026-10-06 to 2026-10-06.
Key findings
- Cheapest 12 GB GPU: RTX 3060 at $0.110/hr, lowest of 4 models in that memory tier.
- Cheapest 16 GB GPU: V100 at $0.088/hr, lowest of 8 models in that memory tier.
- Cheapest 24 GB GPU: RTX A5000 at $0.176/hr, lowest of 7 models in that memory tier.
- Cheapest 32 GB GPU: RTX 4080 Super at $0.338/hr, lowest of 6 models in that memory tier.
- Cheapest 48 GB GPU: RTX A6000 at $0.363/hr, lowest of 6 models in that memory tier.
- Widest price spread: L40S, where the p90 price ($2.91/hr) is 3.4x its lowest ($0.869/hr).
- Most offered GPU: RTX 4090, with 104 live offers from 3 providers.
Methodology
Prices are on-demand list prices, normalized to a per-GPU hourly USD rate using each offer's authoritative GPU count, from the same aggregation that powers the Aquanode GPU Index. A high-side outlier guard drops listings priced far above their model's median; a cheap listing is never suppressed. Spot, interruptible and committed-use pricing are not included.
Lowest is the cheapest per-GPU offer for the model. Data-center lowest is the cheapest offer from data-center (secure) supply only, the same figure shown for that model on pricing, and reads n/a when a model has no data-center offer. Median and p90 are the 50th and 90th percentile across the model's offers.
The report also keeps one snapshot of the index per UTC day. A month's figure is the median of its daily snapshots. A month-over-month change compares monthly medians and is shown only for a model with at least 7 daily snapshots in both months. Days with no snapshot are not filled in.
Cite this report
Plain-text citation
Aquanode GPU Cloud Price Report, October 2026, Aquanode, data through 2026-10-06 18:46:59 UTC. https://www.aquanode.io/gpu-price-reportPage
https://www.aquanode.io/gpu-price-reportJSON: daily history (CORS-open, CC BY 4.0)
https://www.aquanode.io/api/gpu-price-historyJSON: live price index (CORS-open, CC BY 4.0)
https://www.aquanode.io/api/gpu-price-indexLicensed under CC BY 4.0. Live per-model detail is on the Aquanode GPU Index.
