GPU Sidecar — ZephAI Admin

zephai-gpu

Standard_NC4as_T4_v3 · NVIDIA T4 16 GB VRAM · West US 2

Checking...

Compute cost — LLM provider —

◌

VM booting — Azure allocating T4 hardware ~2 min

◌

vLLM loading model — transferring weights to VRAM ~2 min

KV Cache

—

Active Requests

—

Prefix Cache Hit

—

⚠️ Stopping will automatically switch the LLM provider back to Bedrock before shutting down the VM.

Resource Groupzephai-prod-rg

VM SizeStandard_NC4as_T4_v3

Private IP10.0.0.10

vLLM endpointhttp://10.0.0.10:8001/v1

Public IPNone (internal only)

Auto-shutdown00:00 UTC daily

On-demand cost$0.53/hr compute + ~$0.03/hr disk

Connecting...

🖥️

Infrastructure

Azure resources · auto-refreshes every 30s

⚡

App VM — zephai-vm

Hosts ZephAI API, worker, nginx, Redis, ChromaDB

🐘

PostgreSQL — zephai-postgres

Flexible server · stop to pause compute billing

🖥️ GPU Sidecar