— TARX DESKTOP · LOCAL-FIRST AI
FREE TO
THINK
ON YOUR
COMPUTER.
TARX runs AI privately on Apple Silicon. Free local runtime. More compute when you ask.
WHAT TARX
CAN DO
LOCAL AI RUNTIME
Runs on your Apple Silicon machine first. Your work starts on hardware you control.
PRIVATE MEMORY
TARX keeps project context, decisions, and working history close to the user.
SKILLS & TOOLS
Run useful workflows from your desktop with permission, review, and control.
SUPERCOMPUTER HEADROOM
When local compute is not enough, TARX can boost into distributed capacity.
BOOST IN
THREE STATES
TARX can use idle compute when your machine has capacity to spare.
SUPERCOMPUTER NOTES →1tarx.supercomputer.status23local_runtime: ready4contribution: available5headroom: ready6mode: local_first7state: in_the_green
A GRID OF
REAL COMPUTERS
STAY IN
THE GREEN
SKILLS FOR
REAL WORK
PRIVATE BY
DESIGN
LOCAL FIRST
Sensitive work starts on your machine, not someone else's cloud.
PERMISSION SCOPES
TARX is built around user approval, scoped access, and reviewable actions.
MEMORY WITH BOUNDARIES
Working context belongs in Memory. Credentials and secrets require a harder wall.
READY FOR GOVERNANCE
TARX is designed for teams that need privacy, auditability, and rollout control.
BUILT FOR
BUILDERS
TARX gives builders a local runtime for AI workflows, skills, memory, and tools — starting on the desktop, with a path to Supercomputer headroom.
Start on the machine you already use.
Run useful workflows with permission and review.
Carry context across projects and sessions.
Scale heavier jobs when local compute is not enough.
1tarx install2tarx status34runtime: local5platform: apple_silicon6supercomputer: ready7windows: coming
“Stop sending every idea, repo, and prompt into someone else's cloud.”
FREE TO THINK.
BOOST WHEN NEEDED.
COMPUTER
Local AI on your machine.
- +Apple Silicon at launch
- +Private by default
- +Local runtime
- +Memory and skills
- +Windows coming
SUPERCOMPUTER
More headroom when local compute is not enough.
- +Starter: 10M Joules
- +Pro: 30M Joules
- +Max: 150M Joules
- +Priority headroom
- +Voice and vision
ENTERPRISE
Private runtime. Your hardware. Your governance.
- +Private Supercomputer
- +BYO provider path
- +Governance controls
- +Rollout control
- +Custom support
LOCAL AI IS FREE TO RUN. PAID PLANS ADD SUPERCOMPUTER HEADROOM.
START LOCAL.
THINK
FREE.
Download TARX for Apple Silicon. Run AI on your own machine, then boost into more compute when you need it.