ownership · today · What is an on-prem alternative to ChatGPT or Claude seats?
Private AI on the premise.
Join TARX-OS on computers you already own. $0 seats. Get a T-4 or T-70 when the work needs dedicated NVIDIA. Data stays on the premise unless you approve a route. Typical Claude mix is 80% $20 and 20% $200, whole seats. TARX-OS seats are $0. Hardware is list ÷ 36.
Typical Claude mix is 80% $20 and 20% $200, whole seats. TARX-OS seats are $0. Hardware is list ÷ 36.
TARX-LOCAL handles private work on your computer. Connections stay off until you allow them.
Product proof. Not atmosphere. TARX-OS on silicon you own.
HOW TARX WORKS
How TARX works in the enterprise
The company already has Macs. Put TARX-OS on them. That is Join. Prompts, files, and weights stay inside unless someone approves a route. Policy, approval, and Evidence are the same contract the CIO can point at.
When a department needs dedicated NVIDIA, get a T-4 beside the team. When the room needs rack capacity, get a T-70 in 1U. You are not buying another chat subscription. You are adding a Supercomputer node.
This is the sector article. Compare seats to silicon with whole numbers. Five people on a typical mix: $280/mo. A T-4 is $6,600 ÷ 36 = $183/mo. A T-70 is $16,000 ÷ 36 = $444/mo. Air-gapped capable. Evaluation terms in Deployment Design.
THE STACK
Join the Supercomputer. Or own Private Hardware.
TARX-OS on computers you already own. Supercomputer hardware when the site needs dedicated NVIDIA.
Placeholder
TARX-OS · Conversation into the runtime · $0 seats
Prompts, files, and weights stay inside the company unless you approve a route.
A private cluster the CIO can point at
Macs already running TARX-OS. T-4 beside the team. T-70 in the rack they already lock.
Audit after the action
What ran, what changed, who approved. Same contract on TARX-OS and Supercomputer.
KEY FEATURES
Dialed to this question.
Starts onExisting Apple silicon · $0 seats
Desk capacityT-4 · indoor / vehicle
Rack capacityT-70 · 1U rack
Model pathTARX-OS on Mac. T-Supercomputer on the mesh.
GovernancePolicy. Approval. Evidence.
HOW TO START
Four steps. Same contract.
01
Join on computers you already secure
TARX-OS. $0 seats. First-run stays local.
02
Set Control
Writes wait. Search can stay off.
03
Add a T-4 at the department
When shared work needs dedicated NVIDIA.
04
Add a T-70 in the rack
When the room needs rack capacity. Terms in Deployment Design.
NUMBERS
Seats versus silicon you own (five people, 36-month hardware)
Path
Monthly
What you get
Typical Claude mix (4×$20 + 1×$200)
$280
Whole seats. Rented API.
All-Pro floor (5×$20)
$100
Still seats. Still a hop.
TARX-OS on existing Macs
$0
Join. Local runtime. Approval. Evidence.
T-4 Supercomputer
$183
Dedicated NVIDIA on the desk. Same OS.
T-70 Supercomputer
$444
1U rack. Same OS.
FAQ
Questions this page owns.
Is this a new SaaS seat?
No. Join the Supercomputer on TARX-OS is $0 seats. Get a Supercomputer when you want dedicated NVIDIA on the premise.
How does that compare to a typical Claude mix?
Typical mix is 80% $20 Pro and 20% $200 Max, whole seats. Five people: 4 × $20 + 1 × $200 = $280/mo. T-4 is $6,600 ÷ 36 = $183/mo. T-70 is $16,000 ÷ 36 = $444/mo.
Where does the data go?
It stays on the premise unless you approve a route. Air-gapped capable.
Do we need NVIDIA to start?
No. Start on Apple silicon you already own.
TWO EQUAL PATHS
JOIN THE SUPERCOMPUTER. OR OWN PRIVATE HARDWARE.
Join the Supercomputer. Or own Private Hardware.
JOIN · TARX-OS
Join the Supercomputer
Always-on agents on silicon you own.
TARX is not a chat app. TARX-OS is the always-on governed runtime for AI agents and work. Install it on the Mac you already have. Chat is how you talk to it. The runtime stays on — goals, skills, approval, Evidence.
T-1, T-4, T-7, T-70, T-600. Model numbers. They join the Supercomputer — the mesh — for private cluster, burst, or air-gapped sites. Same OS. Same approval. Same Evidence.
