Rack Systems

Neocloud for AI teams. Real racks, live telemetry, liquid hours.

Built on Robinhood.

[Rack Systems @rack_systems]

Reserved means the hours are held for your team and nobody else runs on them.
Batch fills what reserved leaves idle, and is preempted first. Pricing and SLA in full: https://t.co/L1gXTk3olp

[Rack Systems @rack_systems]

On demand 2.49 · reserved 1.89 · batch 1.19 USD per GPU hour.

Billed per minute. No egress, no storage line items, no support tier.

Three plans, one number each: https://t.co/ibuVDKdO41

[Rack Systems @rack_systems]

The docs are up.

The API, the scheduler, the telemetry feed and the SLA behind C1, written for the people who will run the jobs.

Quickstart is five minutes once you have a key: check capacity, submit a job, read its telemetry with curl or Python.

https://t.co/h7E8TlysED

[Rack Systems @rack_systems]

The docs are up.

The API, the scheduler, the telemetry feed and the SLA behind C1, written for the people who will run the jobs.

Quickstart is five minutes once you have a key: check capacity, submit a job, read its telemetry with curl or Python.

https://t.co/h7E8TlysED

[Rack Systems @rack_systems]

Introducing Rack Systems.

Racks you can hold by the hour.

https://t.co/tCO3rIZpMm

[Rack Systems @rack_systems]

Rack Systems is a neocloud where every hour of hardware is an asset you can hold, spend or sell.
This is the architecture behind it. https://t.co/6FeIZHwQD4

[Rack Systems @rack_systems]

Introducing Rack Systems.

GPU compute should have a spot price. We're building the neocloud that gives it one: real racks, live telemetry, hours you can hold, spend or sell.

More soon.