⚡ Up to 40% cheaper than RunPod & Lambda

Simple, transparent pricing

Per-second billing. SSH access included. Pay in USDC or save 25% with $NEURAL.

GPU Rental

Rent raw GPU compute by the second. Full SSH access. Bring any workload — training, inference, rendering, mining.

RTX 3060
12 GB VRAM · Ampere
$0.09/hr
vs RunPod $0.22/hr  ·  59% cheaper
  • Full SSH access
  • Per-second billing
  • 100 GB SSD storage
  • Pre-installed CUDA 12
RTX 4080
16 GB VRAM · Ada Lovelace
$0.29/hr
vs RunPod $0.44/hr  ·  34% cheaper
  • Full SSH access
  • Per-second billing
  • 200 GB SSD storage
  • Pre-installed CUDA 12 + PyTorch
RTX 3090
24 GB VRAM · Ampere
$0.29/hr
vs RunPod $0.44/hr  ·  34% cheaper
  • Full SSH access
  • Per-second billing
  • 500 GB NVMe storage
  • Pre-installed CUDA 12 + PyTorch
A6000
48 GB VRAM · Ampere
$0.69/hr
vs Lambda $1.10/hr  ·  37% cheaper
  • Full SSH access
  • Per-second billing
  • 1 TB NVMe storage
  • Ideal for large model fine-tuning
Pay in $NEURAL
Any GPU above
25% discount
e.g. 4090 → $0.29/hr in $NEURAL
  • Same hardware, same SSH access
  • 25% off all plans
  • $NEURAL burned on payment
  • Hold $NEURAL → earn from fees
💡 Per-second billing — you only pay for what you use. Spin up a 4090 for 5 minutes and pay $0.033. No minimums, no reservation required.   |   💡 SSH access — connect directly to your rented GPU. Run any Docker image, install packages, bring your own code.

Common questions

How does SSH access work?

After you rent a GPU, we provision a connection to the provider's machine. You get an SSH command to paste into your terminal — connect and you're on a raw Ubuntu machine with your GPU ready to use. No containers, no sandboxes.

How is billing calculated?

Per second. If you rent a 4090 for 37 minutes and 14 seconds, you pay exactly that. No hourly rounding, no minimums. Payments are deducted from your prepaid credit balance.

What if the provider goes offline mid-rental?

Billing stops immediately. If the disconnection is not caused by you, you receive a credit for the lost time. We track provider uptime via heartbeat and flag flaky machines.

What is the $NEURAL discount?

Paying in $NEURAL gives you 25% off all plans. $NEURAL is NeuralPool's native token — mined by the GPU providers powering this network. You can buy it on launch day or mine it yourself by running a provider node.

Is this the same as RunPod or Vast.ai?

Similar model — peer-to-peer GPU rental. The differences: we're cheaper, providers earn $NEURAL on top of their rental income so they're incentivized to stay online, and customers can pay in $NEURAL for a 25% discount.

Start building today

Get an API key in 60 seconds. No credit card required to try the inference API.

Get API Key →