TARX is the provider. The product is T-Supercomputer. Public API https://api.tarx.com/v1. Model t-supercomputer. Provider path tarx/t-supercomputer. TARX-LOCAL stays private on the computer. Howdy is the console, not the inference proxy.
Base URL https://api.tarx.com/v1. Credential: TARX API key (TARX_API_KEY) from the Howdy console preview. Production Supercomputer capacity is being qualified. Commercial pricing is not approved. This is not an “available now” inference SLA.
curl https://api.tarx.com/v1/chat/completions \
-H "Authorization: Bearer $TARX_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"t-supercomputer","messages":[{"role":"user","content":"Hello"}]}'T-Supercomputer is designed to receive demand from compatible developer APIs and inference marketplaces. No marketplace listing, billing, or production routing is live.
Package reusable tool behavior.
Connect TARX-OS context and tools.
Request shapes for local execution.
Install, inspect, and manage TARX-OS.
Authoritative technical guidance.
Open developer kit. Not a production enclosure.
TARX-OS is the developer contract for private AI workflows, skills, memory, and approved tools. Join the Supercomputer. Or own Private Hardware. Same approval.
Start on the machine you already use.
Run useful workflows with permission and review.
Carry context across projects and sessions.
Run on the computer you already own. Add Private Hardware when you need dedicated power.
1tarx install2tarx status34runtime: local5platform: local6supercomputer: validation7windows: coming
No ads. No extraction.