Virtual GPU Nodes
Register and manage rented cloud GPU machines (Spheron, AWS, RunPod, etc.)
Total Nodes
0
Online
0
Inference Ready
0
Total VRAM
0GB
Hourly Cost
$0.00/hr
Endpoint Setup Wizard
Follow these 3 steps in order to get your GPU nodes serving inference. Each step builds on the previous one.
Open Jupyter terminal & install vLLM
SSH is disabled on these instances, but Jupyter is running. Open the Jupyter URL, use its built-in terminal, and run the install command. vLLM runs on port 10100 (Jupyter uses 8080).
Fetch public endpoints from Vast.ai
Reads the public IP + port mapping for each instance. The endpoint URL is auto-saved to each node record.
Probe endpoints to confirm vLLM is live
Hits /v1/models on each node. If vLLM is up, the detected model is saved to the node and it becomes 'Inference Ready'. Run this ~5 min after the SSH install.
No virtual GPU nodes registered yet.
Click "Register Node" to add a rented Spheron GPU machine.