Autonomous GPU Spot Cockpit
Multi-cloud broker telemetry, resident VRAM watchdog, and zero-data-loss spot orchestration.
AWS EC2 Spot Compute
Offline
No Active Box
Public IP:
--
Type & Region:
g6e.xlarge · us-east-1
GPU & VRAM Resident
Offline
0.0 / 48.0 GiB
Device:
NVIDIA L40S
Live Billing Odometer
$0.75 / HR
$0.00
Session Uptime:
0.0 mins
Auto-Shutdown:
20m Idle Guard
Remote Worker Stream (/scratch/worker/worker.log)
[Cockpit] Initializing live telemetry and log stream...
Host Device Compute & Model Registry
Zero-cloud local inference engine. Automatically probed hardware and curated model weights.
Apple Metal (MPS)
probing…
Device Architecture
probing…
Total / Free Memory
probing…
Instruction Set
ARM_Neon_MPS
Local Cache Footprint
~/.cache/spacepilot/models/
Recommended Model Catalogue for Your Device
| Task Category | Model & Precision | Disk Size | Hardware Fit | Action |
|---|
Compute Provider
Where a launched instance comes from, and the credentials it uses.
AWS Spot Fleet Settings
Generation & Guidance Defaults