api.tarx.com is live.
It's an OpenAI-compatible inference API. Keep the request shape, change the execution route.
from openai import OpenAI
client = OpenAI(
base_url="https://api.tarx.com/v1",
api_key="your-tarx-key"
)
response = client.chat.completions.create(
model="tarx-mind",
messages=[{"role": "user", "content": "Hello"}]
)
Same SDK. Same response format. Same streaming. The difference is that TARX is built around route control.
Local work can stay on the user's computer. Hosted Supercomputer work should be approved, attributable, and evidenced.
The public product truth is simple: TARX is free and private on your Computer. Supercomputer is permissioned power when the Computer is not enough.
For Computer development, start with TARX Free. For external integrations, create a developer identity key. For Supercomputer work, use earned access, approved grants, or Power Bundles.
If you're running TARX locally, point at localhost:11435/v1 and the app contract is the same, except the first route is your own machine.
Tool integration is one command:
tarx mcp add claude
That wires TARX tools into compatible clients. Memory, file search, and inference can become route-aware tool calls.
The API is live. Go build something.
