Pre-configured clusters shipping now

Your AI Cluster. Delivered Ready.

Pre-configured Apple Silicon clusters with autonomous ops built in. Plug in, power on, run inference. No Linux. No cloud bills. No setup.

500Wtotal drawvs 3,000W GPU
2.2yrbreakevenvs cloud
Zerosetup timeplug in, run

Choose your cluster

Every tier ships burn-tested, ABM-enrolled, and ready to run inference on day one.

Start here

Single Unit

from $2,499one-time
  • Mac Mini, Mac Studio, or MacBook Pro
  • r1o stack pre-loaded
  • 72-hour burn-tested
  • ABM enrolled
  • 1 model pre-loaded
  • Add more units anytime
Best valueMost popular

Pro Cluster

$15,999one-time
  • 4x Mac Mini M4 Pro 64 GB
  • 256 GB total unified memory
  • 3x TB5 cables included
  • RDMA mesh + JACCL distributed inference
  • r1o Agent Stack + asmi monitoring
  • 3 models pre-loaded and serving
  • 72-hour burn-in per unit
2,456 GB/s bandwidth

Max Cluster

$27,999one-time
  • 4x MacBook Pro M5 Max 128 GB
  • 512 GB total unified memory
  • 3x TB5 cables included
  • RDMA mesh + JACCL distributed inference
  • r1o Agent Stack + asmi monitoring
  • 5 models pre-loaded and serving
  • Portable — take your cluster anywhere
Maximum inference

Ultra Cluster

$42,999one-time
  • 4x Mac Studio M3 Ultra 256 GB
  • 1 TB total unified memory
  • 6x TB5 cables included
  • RDMA mesh + JACCL distributed inference
  • r1o Agent Stack + asmi monitoring
  • Unlimited models pre-loaded
  • 90-day post-deploy support
2 TB unified memory

Sovereign

$89,999one-time
  • 4x Mac Studio M3 Ultra 512 GB
  • 2 TB total unified memory
  • 6x TB5 cables included
  • Runs 400B+ parameter models natively
  • r1o Agent Stack + full automation
  • White-glove on-site installation
  • 1-year dedicated support
  • Discontinued config — sourced exclusively

What's included

FeatureStarterProMaxUltraSovereign
Hardware1 unit4 units4 units4 units4 units
Total unified memory128 GB256 GB512 GB1 TB2 TB
Memory bandwidth546 GB/s273 GB/s ×4614 GB/s ×4800 GB/s ×4800 GB/s ×4
TB5 cables included3x3x6x6x
TB5 RDMA
r1o Stack
Agent StackAdd-on
Models pre-loaded135UnlimitedUnlimited
400B+ models
Portable
SupportEmail30 days30 days90 days1 year
On-site setupAvailable

Why not cloud?

The math is straightforward. Own your compute.

Cloud GPU

  • $18,400 / year recurring
  • Data leaves your network
  • Scaling = more bills
  • NVIDIA dependency & allocation lottery

RackStudio

  • One-time $13,999
  • Data stays on-prem, always
  • Scales with more units
  • Your hardware, forever
Included with Pro & Ultra

The r1o Agent Stack

Your cluster's brain. Autonomous operations that monitor, heal, and manage your inference fleet so you don't have to.

Apple Business MCP

Device enrollment, MDM, compliance

Cluster Intelligence

Real-time GPU/RAM/power monitoring via asmi

Self-Healing

Auto-restart failed inference, recovery agents

Model Management

Deploy, quantize, shard across units

Hermes Notifications

iMessage/Slack alerts for health events

NanoMDM

Push profiles, manage your fleet remotely

Ready to own your inference?

Reach out and we will scope your cluster, ship it configured, and get you running inference the day it arrives.