Become an Operator

Run a Tangle Blueprint, serve model inference, and earn a share of every request routed to you. Pay out on-chain, no invoicing.

Get started in three steps

1. Pick a Blueprint

vLLM, Ollama, TGI, custom — any deployable inference target.

2. Stake TNT

TNT collateral required to register on-chain. Slashable on SLA violations.

3. Register

cargo tangle blueprint deploy — live on testnet in minutes.

What you'll need

  • ·A GPU box (NVIDIA A100/H100) or a Modal/cloud account
  • ·An OpenAI-compatible endpoint exposing /v1/chat/completions and /health
  • ·TNT to stake as on-chain collateral
  • ·The Tangle CLI (cargo tangle) to deploy your Blueprint