Every invocation, with retries, latency, and live tail.
Success rate, p50/p95/p99, per-endpoint sparklines. No external observability stack to wire up.
لائیو فی گھنٹہ قیمتیں
Auto-bills every hour while running. Stop the pod to pause; terminate to release the GPU.
| GPU | VRAM | Per hour |
|---|---|---|
| RTX A4000 | 16 GB | $0.34 |
| RTX 4090 | 24 GB | $0.68 |
| L40S | 48 GB | $1.58 |
| A100 80 GB | 80 GB | $2.78 |
| H100 80 GB | 80 GB | $5.38 |
| H200 | 141 GB | $7.18 |
Prices shown are list. Live availability and any spot discounts are surfaced at request time on /infra/pods/new.
ٹیمیں Hypereal کو کیوں منتخب کرتی ہیں
SSH in 60 seconds
Provision a pod, copy the SSH command, and you are in. PyTorch, CUDA, and Jupyter pre-installed on every image.
Persistent volumes survive Stop
Stop your pod to pause the meter without losing your weights, datasets, or virtual env. Resume anytime.
Auto-bill hourly, terminate to stop the meter
Billed per hour while running. Terminate to fully release the GPU and stop all charges. No minimum commit.
یہ کیسے کام کرتا ہے
Pick a GPU
Choose from A4000 up to H200 — live availability shown at request time.
Set image (PyTorch default)
Default PyTorch 2.x + CUDA image, or paste any public Docker image URL.
Get SSH host + key
Your private key is generated on creation. Copy the one-line SSH command and connect.
Stop or Terminate anytime
Stop to pause billing while keeping the disk. Terminate to fully release.