Serve open models on your GPUs. Earn per token.
You have idle H100s, H200s or B200s, or spare capacity on DeepInfra, Together, Fireworks or Novita. Route our requests for open-weight models and we pay you for each one.
Apply as a provider
Your hardware and keys stay yours
You keep your accounts and your machines. We route requests to you, we encrypt any credential you share before we store it, and our operators see its first eight characters and nothing more.
We pay per request
Each request we route to you adds to your balance, and we pay the full balance out in one transfer.
You see the numbers
Your dashboard lists earnings per model, the amount still owed, and the request counts behind both.
Tell us what you can supply
You create no account and commit to nothing by sending it.
- We read each application in the order it arrived.
- We reply by email, whatever the answer.
- If we accept, we route requests to you and pay for each one.