This page is the high-level checklist. For the full setup, branch into one of:

Standard provider (full)

Run a proxy-router on your own infrastructure, point it at your model.

TEE provider (SecretVM)

Deploy the hardened -tee image on SecretVM and tag your model tee.

Resale provider

Resell capacity from Venice / OpenAI / Anthropic — including Venice Diem holders monetizing API access.

Docker

Containerized proxy-router only.

AWS EC2

Two-instance EC2 setup (LLM + proxy-router).

Checklist

1

Have a backend model

OpenAI-compatible endpoint reachable from your proxy-router host (e.g. http://my-model:8080/v1/chat/completions).
2

Look up what already exists

Before minting a new marketplace model, check active.mor.org/status (or active_models.json / active_bids.json) for the model name you intend to serve and competing prices. Prefer bidding on an existing modelId. See Register on chain → Step 0.
3

Have a funded wallet

BASE wallet with enough MOR for provider stake + bid fee (and model stake only if you must mint a new model), plus a little ETH for gas. See Networks and tokens.
4

Have a publicly reachable endpoint

The proxy-router needs to be reachable on host:3333 from the internet so consumers can connect. The Swagger/admin UI on :8082 should not be public — see API auth. SecretVM exposes HTTPS admin on :443, which pairs well with MyProvider.
5

Run the proxy-router

Configure .env, models-config.json, rating-config.json and start the binary — or follow SecretVM quickstart. See Provider quickstart.
6

Register on chain (prefer MyProvider)

Use MyProvider when you can (HTTPS node / SecretVM): register the provider, then bid on an existing model. Mint a new model only when nothing suitable exists (or you need a new tee-tagged listing). Swagger/API steps: Register on chain. Every postModelBid charges a non-refundable 0.3 MOR fee — finish local config before your first bid; see Pricing.
7

Verify end-to-end

Don’t trust startup logs alone — run the five-step self-check in Verify your provider setup: healthcheck, public TCP, on-chain records, active.mor.org discovery, and a real prompt via the hosted Inference API.
TEE providers must register with the model tag "tee" for any of the two-hop attestation chain to engage on the consumer side. See TEE overview.