Pods that come and go
Surfra treats pods like serverless replicas. Queued work starts them, idle ones stop after your timeout.
RunPod on Surfra
Your RunPod account, autoscaled.
RunPod has one of the widest GPU menus around. Surfra starts pods in your account when work is waiting, stops them when it goes quiet, and can fall back to another cloud when the card you want is sold out.
Connect
What you need from RunPod.
Add your RunPod account under Credentials. Surfra checks the key with RunPod before saving it, then encrypts it with your team's key. It is never shown again.
| Detail | RunPod |
|---|---|
| You provide | API key |
| Who bills you | RunPod, directly, at its own price |
| Surfra markup | None |
| CPU machines | Supported |
GPUs
RunPod GPUs you can rank.
Surfra can place these GPUs on RunPod, wherever your account has capacity. Prices and stock come from RunPod.
- RTX 3090
- RTX 4090
- RTX 5090
- RTX A4000
- RTX A5000
- RTX A6000
- RTX 4000 Ada
- RTX 5000 Ada
- RTX 6000 Ada
- RTX PRO 6000
- L4
- L40
- L40S
- A40
- A100 40GB
- A100 80GB
- H100
- H200
- B200
How it runs
What Surfra adds on top of RunPod.
One queue across clouds
Put RunPod next to Salad, Vast.ai, or your own machine. Your app keeps one name, one queue, and one set of logs.
RunPod's price, unchanged
Pods start in your account and RunPod bills you at its own rate.
Other clouds
Pair it with a second cloud.
FAQ
Common questions
Is this the same as RunPod Serverless?
No. Surfra starts regular pods in your RunPod account and runs the queue and scaling itself, so the same app can also use other clouds.
Do I pay Surfra for RunPod time?
No. RunPod bills you directly. Surfra adds no markup.
Autoscale RunPod from your own account.
Surfra is invite only while we test it. Tell us what you want to run and we will send an invite.
Request access