NVIDIA RTX 4090 cloud GPU prices
Compare listed RTX 4090 rentals for development and workloads that fit within the reported GPU memory. The price table covers our tracked providers, not every marketplace listing.
Rates dated Nov 10, 2026–Nov 10, 2026
Publication dates, not live quotes. Confirm the current rate and capacity with the provider.
Catalog generated .
Lowest listed on-demand rate
$0.450/GPU·hr
CloudRift · RTX 4090 · Unspecified · 24 GB per GPU · 1 GPUs · $0.450/instance·hr · ap-east-tw-kn-2
Lowest listed spot / interruptible rate
No listed offer
Compare rental prices
Lowest listed compute rates. Explore comparison filters for your region.
| Provider / GPU / variant | VRAM per GPU | Pricing | Per GPU / hr | Full instance / hr | Region | Provider / rate date |
|---|---|---|---|---|---|---|
| CloudRift RTX 4090 · Unspecified Instance: rtx49-18-53-450-kn-packed.1 Link to offer | 24 GB | On-demand | $0.450 | $0.450 1 GPUs | ap-east-tw-kn-2 Stock unknown Stock & project cost | Check provider priceRate dated |
| CloudRift RTX 4090 · Unspecified Instance: rtx49-18-53-450-kn-packed.2 Link to offer | 24 GB | On-demand | $0.450 | $0.900 2 GPUs | ap-east-tw-kn-2 Stock unknown Stock & project cost | Check provider priceRate dated |
| CloudRift RTX 4090 · Unspecified Instance: rtx49-18-53-450-kn-packed.3 Link to offer | 24 GB | On-demand | $0.450 | $1.35 3 GPUs | ap-east-tw-kn-2 Stock unknown Stock & project cost | Check provider priceRate dated |
| CloudRift RTX 4090 · Unspecified Instance: rtx49-18-53-450-kn-packed.4 Link to offer | 24 GB | On-demand | $0.450 | $1.80 4 GPUs | ap-east-tw-kn-2 Stock unknown Stock & project cost | Check provider priceRate dated |
| CloudRift RTX 4090 · Unspecified Instance: rtx49-18-53-450-kn-packed.5 Link to offer | 24 GB | On-demand | $0.450 | $2.25 5 GPUs | ap-east-tw-kn-2 Stock unknown Stock & project cost | Check provider priceRate dated |
| CloudRift RTX 4090 · Unspecified Instance: rtx49-18-53-450-kn-packed.6 Link to offer | 24 GB | On-demand | $0.450 | $2.70 6 GPUs | ap-east-tw-kn-2 Stock unknown Stock & project cost | Check provider priceRate dated |
Manufacturer specifications, with conditions
Reviewed September 26, 2026 against NVIDIA documentation. These are hardware specifications and theoretical peaks, not NVGPU measurements or third-party benchmark results. Sparse Tensor Core peaks require supported structured sparsity; they are not expected dense training throughput.
| Reference variant | Memory | Memory bandwidth | FP16 Tensor Core peak | Conditions and source |
|---|---|---|---|---|
| GeForce RTX 4090 | 24 GB GDDR6X | Not compared here | Not compared here | No NVLink. Memory capacity and software support constrain suitable workloads; no training-speed result is implied. NVIDIA source |
NVGPU has not run application benchmarks for these comparisons. To assess a third-party result, require the model and revision, precision, batch size, sequence lengths, GPU count and variant, software versions, interconnect, test date and methodology.
Before you rent
- Check model memory use before selecting a consumer GPU instance.
- Confirm driver, CUDA and container support for your software.
- Check host resources, storage persistence and provider support before using a rental for production.
What is included?
These are listed compute rates. The instance shown is the configuration being priced, not evidence of the provider's smallest available rental. Storage, data transfer, tax, licenses and support can be additional. Reserved contracts, commitment discounts and serverless per-request charges are not compared. Listed spot capacity can disappear or be interrupted.
Hourly price vs. total cost
Multiply the full-instance hourly rate by billed hours, then add storage, transfer and other charges. Compare the same workload and quality target. Rental throughput and uptime have not been measured by NVGPU.