Provider onboarding

Register once, then attach any mix of nodes.

Providers are account-level actors. A provider profile belongs to a user account, carries payout state and policy, and can later own many worker nodes.

You do not classify nodes here. Register first, choose a supported node setup, run the setup script with the key from the website, then start the worker.

Region setPick one v1 region for the provider profile.
Provider keyReturned immediately for worker authentication.
Node setupUse the recommended Docker lane, run the setup script, then restart the worker.
Create provider

Provider flow

Result will appear here.
The returned provider key is the bootstrap secret for worker registration. After signup, pick the recommended Docker setup, run the setup script, and log back in with the same email/password on the provider dashboard. Model class and pricing are inferred from the approved models the node reports, not from this form.
Node setup

Simple v1 operator flow

  1. Create the provider profile and copy the returned provider key.
  2. Download the provider bootstrap installer from this site.
  3. Run the installer to fetch the worker bundle and Docker files.
  4. Run bash infra/docker/setup-provider-node.sh and paste the values from the website.
  5. Run bash infra/docker/deploy-provider-worker.sh restart.
  6. Open the provider dashboard and confirm the node is online.
curl install command will appear here.
wget install command will appear here.
Today the worker lane is optimized for local Ollama. Larger vLLM-focused hosts can get a separate setup lane as that adapter lands.
Recommended hardware

Fast VRAM matters

Recommended VRAM speed

8GB >400 GB/s

24GB >900 GB/s

For many AI workloads, memory bandwidth matters as much as or more than raw VRAM size. A smaller high-bandwidth GPU may be more valuable than a larger but slower card.

Privacy Node policy

Privacy Node agreement

By marking your node as a Privacy Node, you explicitly agree that you will not inspect, log, store, retain, analyze, train on, share, or otherwise use customer traffic, prompts, outputs, or metadata beyond what is strictly required to process the request and operate the service.

You also agree not to attempt to identify customers, extract sensitive information, profile usage patterns for unrelated purposes, or reuse customer traffic for benchmarking, dataset creation, model training, or commercial advantage.

If temporary processing artifacts, caches, or logs are created as part of normal system operation, you agree to minimize them and not use them for any secondary purpose.

Resonara may remove the Privacy Node classification from any provider that is suspected of violating these commitments.