Cluster Online · 4/4 Compute Engines Operating Normally
UTC Live Telemetry
Active Compute Nodes
4 / 4
Total Draw: ~1,950 W
Unified Memory
384 GB
4x 96 GB NVLink Bandwidth
Inference Throughput
846 tok/s
Gemma-4-31B + Qwen38 NVFP4
Routable Endpoints
19 Services
Zero-Downtime Caddy Mesh
Hardware Core Allocation
GPU 0 51°C · 316W
Visual Witness & Aesthetics
Qwen3.8-27B NVFP4 (:8001) + Pixtral-12B (:8002) + FLUX Living Garden
Memory 94.1 / 96 GB
Compute 99%
GPU 1 55°C · 506W
Council Governor & Reasoning
Gemma-4-31B-IT NVFP4 (:9000 & :8000) + CUDA Shard Vault Daemon
Memory 73.1 / 96 GB
Compute 100%
GPU 2 67°C · 577W
Arcane 3D World & Character Studio
FLUX.1-dev + Kontext (swapped) + CLIP Defect Gate + Anime Studio
Memory 36.4 / 96 GB
Compute 100%
GPU 3 61°C · 558W
Lumen Continuous Light Studies
FLUX.1-dev Continuous Creative Pipeline + Contact Sheet Self-Review
Memory 35.7 / 96 GB
Compute 99%
Status Service Name Endpoint Domain Hardware Allocation Port Latency Actions
GPU 1

Service Details

Architecture Mappings
Public Domain
Internal Upstream
cURL Invocation
curl -sI https://...
Launch Endpoint
Copied to clipboard