AXION Cloud · demo control plane
Console
This is a working control plane over invented resources. Instances take ~22 seconds to provision, report synthesized utilization, and meter per second against an anonymous session cookie. Nothing is provisioned anywhere and nothing is charged.
Session billing
Per-second metering against an in-memory control plane. Resets when the server restarts.
Loading…
Launch capacity
POST /api/v1/instances
1 accelerators · 48 GB VRAM each · 100 GbE · $1.48/hr — $1,065.60/mo
A job label drives the simulated utilization up, as if the node were busy.
Instances
GET /api/v1/instances — telemetry is synthesized, refreshed every 2.5 s
No instances. Launch something above — it provisions for ~22 seconds, then reports telemetry.
Serverless endpoints
POST /api/v1/endpoints — scale-to-zero, per-token metering
Playground
POST /api/v1/chat/completions — OpenAI-compatible shape, hollow inside
Metered per token, so watch the billing panel move.