vLLM

Choose a model and deploy it as a vLLM serverless endpoint. There are three menus by compute type — GPU · CPU · Rebellions — with the same screens and usage.

Menu	Compute	Notes
GPU	GPU	Used for most models.
CPU	CPU	Serves lightweight models without a GPU. The creation screen has an extra Resources (CPU, memory) section.
Rebellions	Rebellions NPU	Serves models prepared for Rebellions. The creation screen has an extra Resources section.

• Browse endpoints
• Create an endpoint
• Manage endpoints (URL, test prompt, logs, edit, delete)